Real-Time Proactive Equipment Operator and Ground Worker Warning and Alert System in Steel Manufacturing

Transcription

1 Real-Time Proactive Equipment Operator and Ground Worker Warning and Alert System in Steel Manufacturing Hazards are ever-present in the steel plant environment, and a heightened awareness and emphasis on safety is a necessary priority for our industry. This monthly column, coordinated by members of the AIST Safety & Health Technology Committee, focuses on procedures and practices to promote a safe working environment for everyone. Authors Eric Marks Ph.D. candidate, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, Ga., USA ericmarks@gatech.edu Jochen Teizer Associate Professor, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, Ga., USA teizer@gatech.edu Contact Comments are welcome. If you have questions about this topic or other safety issues, please contact safetyfirst@aist.org. Please include your full name, company name, mailing address and in all correspondence. This article is the first in a three-month series of Safety First articles featuring the reports from the recipients of the 2011 Don B. Daily Memorial Fund. The steel manufacturing environment is characterized by a mixture of multiple resources, such as personnel, equipment and materials. Each of these resources performs dynamic activities in a defined space, and they often are in close proximity to each other. When steel manufacturing material handling equipment is operating in close proximity to ground workers, a potentially hazardous situation is created. Contact collisions between ground workers and material handling equipment can increase the risk of injuries and fatalities for construction personnel. A lack of scientific evaluation data currently exists for safety technologies, such as proximity detection and warning systems in steel manufacturing. Minimal information and data currently exist to evaluate how safety technologies already in place can be implemented to warn ground personnel of the presence of hazardous proximity situations. Experiments designed to emulate the outdoor steel manufacturing environment need to be developed to evaluate these emerging safety technologies. Data retrieved from these experiments can be used to demonstrate the capability and effectiveness of these safety technologies in the steel manufacturing environment. A review of current steel manufacturing safety statistics, safety best practices and existing safety technologies including proximity sensing and alert technology will be executed in this work. These technologies will be reviewed specifically for their effectiveness to perform in the steel manufacturing environment. Recommendations for future work on this issue, including potential applications to promote safety in steel manufacturing, will be discussed. Literature Review The material staging areas of steel manufacturing environments are characterized by movement and interaction between ground workers and material handling equipment. The dynamic nature, unique size and scope of material staging areas in steel manufacturing environments can create hazardous proximity issues. The following review covers current injury and fatality incidents associated with proximity issues in the steel manufacturing industry. The review will also cover current safety practices of workers in hazardous proximity situations, existing proximity sensing and alert technologies, and previous methods used to evaluate these technologies. A research needs statement is derived from findings of the review. Injury Statistics Related to Workers and Material Handling Equipment The manufacturing industry continues to be one of the most dangerous industry sectors for workers when 56 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

2 Table 1 Occupational Fatalities, Injuries and Illnesses for the Manufacturing Industry 1 Year(s) Fatalities Non-fatal injuries Non-fatal illnesses (9.7%) 1,273,171 (25.0%) 185,818 (51.1%) (7.9%) 630,660 (18.0%) 59,100 (31.3%) (7.0%) 480,700 (15.6%) 47,900 (28.7%) (7.0%) 454,100 (15.6%) 47,000 (30.0%) compared to other industries in the United States. The Bureau of Labor Statistics found that the manufacturing industry experienced the second most injuries and illnesses in Since 1994, the manufacturing industry accounted for 25% of injuries and illnesses. Table 1 shows the number of fatalities, as well as non-fatal injuries and non-fatal illnesses, experienced by the manufacturing industry. Numbers in parentheses are the percentage each value represents when compared to all fatalities, injuries and illnesses experienced by the U.S. private sector. The Bureau of Labor Statistics also records fatalities of specific sectors of the manufacturing industry. 1 The metal manufacturing industry experienced 215 fatalities from 2003 to Ninety-eight of the fatalities during this time period resulted from workers colliding with equipment or objects. Fatalities experienced per year by the metal manufacturing industry are shown in Figure 1. Current Safety Practices Standards and regulations mandated by the Occupational Safety and Health Administration (OSHA) are imperative to promoting safety in construction, but are currently not capable of preventing contact collisions between workers and steel manufacturing equipment. Current safety regulations require passive Figure 1 Metal manufacturing fatalities per year. 1 safety devices such as hard hats, reflective safety vests and other personal protective equipment (PPE). These passive safety devices are incapable of alerting construction operators and workers in real time during a hazardous proximity situation. Other safety regulations, such as safety training and education, can increase the awareness of close proximity issues for equipment operators and workers. The National Institute for Occupational Safety and Health (NIOSH) conducted a review of fatalities involving forklifts from 1980 to Results of the review indicated that 20% of the total forklift fatalities involved a worker on foot colliding with a forklift. This category of fatality accounted for the second highest portion of the forklift-related fatalities reviewed. Several regulations have been implemented by OSHA to address proximity problems and resulting fatalities between ground workers and forklifts. Three specific OSHA regulations pertain to forklift operations in steel manufacturing material handling areas. The following regulations are an attempt to combat hazardous proximity issues in these environments: 7 Under all travel conditions, the forklift shall be operated at a speed that will permit it to be brought safely to a stop. The operator shall slow down and sound the horn at cross aisles and other locations where vision is obstructed. Forklifts shall not be driven up to anyone standing in front of a bench or other fixed objects. Equipment Operator Visibility Most material handling environments in steel manufacturing are characterized by a multitude of interactions between heavy equipment, workers and materials. This dynamic environment creates many visibility issues. Blind spots impede the sight of equipment operators, creating non-visible areas for the operator outside of the equipment cab. Contact collisions can occur when workers, equipment or materials enter these blind spots undetected by the equipment operator. The close proximity of workers on the ground to heavy equipment creates many visibility-related issues for equipment operators. Blind spots restrict an operator s line of sight to objects outside of the equipment cabin. Many accident investigations and research studies found visibility-related issues created by blind spots caused operators to: (1) run over workers and materials, (2) contact other equipment and (3) roll over equipment. 2 Diagrams of equipment operators blind areas were compiled by NIOSH for 43 pieces of heavy equipment. These blind area diagrams were created using manual methods prescribed in the International Standards Organization (ISO) codes 5006 (Earthmoving machinery operator s field AIST.org October

3 of view test method and performance criteria) and (Earth-moving machinery field of vision of surveillance and rear-view mirrors) standards. 4 Real-Time Proactive Proximity Warning and Alert Technology The material handling environments of steel manufacturing and the construction environment have many similar characteristics. Both consist of ground workers, equipment and materials all performing dynamic tasks in a harsh outdoor environment. Because of the similar conditions, both environments experience hazardous proximity issues between ground workers and equipment. A study performed in 2007 found that real-time safety technologies implemented on construction jobsites are capable of providing alerts to construction workers and equipment operators in real time when a hazardous proximity issue is present. 13 These technologies can create a safety barrier and provide workers with a second chance if other safety best practices are disregarded. 16 Some of these technologies are also capable of recording safety data that are currently not obtainable, such as close calls or nearmisses. This new information and warning system can present new data sources and potentially improve safety in construction. In 2001, T.M. Ruff found several proximity warning systems including radio detection and ranging (radar), sonar, global positioning system (GPS), radio transceiver tags, cameras, and combinations of the mentioned technologies. The study found each of the candidate technologies to have limitations in the construction environment, such as availability of signal, size, weight and feasibility. 9 A few similar studies also investigated candidate technologies to combat hazardous proximity issues. Technologies excluded from the study were those still in the prototype stage and not yet commercially available. 15 The evaluated technologies for this study included radio frequency identification (RFID), ultrawide band (UWB), GPS, magnetic marking fields, vision detection devices including video cameras, sonar, laser and radar-based proximity warning technologies. Several parameters such as read range, ability to calibrate the system, alert method, precision, capability of performing in an outdoor environment and others were used to assess each candidate technology. 15 Table 2 shows the benefits and limitations of select proximity warning and alert technologies. Radio frequency technology can: Decrease the risk of collisions. Provide alerts in real time for equipment operators. Create a tool for managing risk. Monitor with minimal distractions (e.g., nuisance alerts) during normal operation. Create an additional protection layer for ground workers. Capable of performing in most outdoor environments. This study also found these limitations to radio frequency technology: System requires a power source. Must be installed on workers and material handling equipment. Equipment components and outdoor conditions have an impact on proximity sensing. Table 2 Benefits and Limitations of Candidate Proximity Detection Technologies 5,9,15 Technology Benefits Limitations GPS Laser Magnetic marking fields Radar Sonar UWB Vision Minimal required infrastructure Low initial cost Functional outdoors High accuracy of data High signal update rate Capable of functioning outdoors Minimal required infrastructure Proximity ranges can be varied System can identify a ground worker from other objects Capable of multiple antenna integration Can be used to supplement video Minimal infrastructure required Low initial cost System can identify a ground worker from other objects 3-D location values in real time System can identify a ground worker from other objects Capable of detection at large/small ranges Not functional indoors Not suited for short range detection High initial cost Not able to identify a ground worker from other objects Accurate only for short range detection Requires a system-specific battery for power source Two devices are required to be installed on a worker Not functional in fast-moving scenarios Not able to identify a ground worker from other objects No long-range detection Minimal detection range Susceptible to outdoor elements Sizeable amount of infrastructure High initial cost Can function outdoors Poor visibility at night or in dusty areas Line-of-sight segmentation High data processing effort 58 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

4 Methods for Testing Proximity Warning and Alert Systems Several past researchers have incorporated methods to evaluate the capabilities of proximity detection and alert systems. A camera and radar systems were mounted on a large-capacity haul truck, and proximity alert distances were manually marked and measured on the ground. 11 Trials included several activities typical of a large-capacity haul truck in copper mining environments. The system detected obstructions approximately 30 feet in front of and behind the vehicle. A similar experiment deployed GPS systems on largecapacity haul trucks, and a base station was located on a nearby hill. 10 Several trials were performed to test the accuracy of the system to track three mobile vehicles and six stationary objects. The system was able to track all vehicles and objects while the haul truck performed typical activities in a surface mining environment. Other research integrated computer-assisted stereo vision with the previously discussed radar detection system to potentially realize the benefits of a combined proximity detection and alert system. 12 In this collaborative research effort with NIOSH, stereo cameras were mounted on the rear of an off-highway dump truck. Several field trials positioned a person and berm in the path of the truck to evaluate the system s proximity detection capabilities. Objective and Methodology to Evaluate Proximity Warning and Alert Technologies The primary objective of this research is to evaluate the capability of real-time proactive proximity sensing and warning devices in steel manufacturing during material handling operations. When material handling equipment and ground workers are in close proximity to one another, the sensing technology will detect the hazardous condition and activate an alarm to warn equipment operators through devices called Equipment and Personal Protection Units (EPU and PPU, respectfully). The research is limited to proximity issues between material handling equipment and ground workers located within outdoor steel manufacturing environments. The radio frequency technology will be evaluated through several experiments. Each experiment is designed to measure the performance of the technology in a simulated and actual outdoor material handling environment. Experiments were designed and executed to test the device s ability to detect and alert equipment operators of hazardous proximity issues while subjected to a simulated materials handling environment. The first experiment tested the proximity sensing devices in a mobile ground worker and static forklift scenario. The PPU was attached to a ground worker outside of the forklift and one EPU was placed inside the forklift cabin. The ground worker equipped with the PPU device approached the forklift at a specified distance from many different angles. A theoretical safety zone was created by plotting the points at which the alert was activated. Similar to the previous experiment, a static ground worker was equipped with a PPU device in this setting. An EPU device was installed on one piece of equipment. For each experiment, the piece of equipment approached the static ground worker at a constant speed. After the alert was activated, the equipment operator halted the piece of equipment and the distance between the equipment s stopped location and the static ground worker was measured. Because the proximity distance was measured after the equipment was stopped, the data represented the minimal distance required to stop the equipment before it struck the ground worker. A final set of experiments was designed to measure the visibility of an operator by utilizing a commercially available laser scanner capable of producing a three-dimensional (3D) spatial point cloud from inside the equipment cabin. Data recorded from the blind spot measurements was analyzed for percent visibility and visible areas versus nonvisible areas of an equipment operator as specified by the ISO standards. Each of the experiments was designed to evaluate a specific characteristic of the outdoor material handling environment. The methods used for measuring proximity alert distances and data collection were held constant for all experiments. The research team reviewed all relevant OSHA safety regulations for equipment operation. The team followed all OSHA safety regulations while conducting experiments in active manufacturing environments, and only qualified equipment operators were used for testing. Experiments and Results The experiments were designed to evaluate the effectiveness of proximity detection technology in the steel manufacturing environment, specifically between materials handling equipment and ground workers. Each experiment attempted to simulate functions characteristic of a typical material handling environment. The proximity detection system utilized for the experiments used a secure wireless communication line of ultrahigh frequency (UHF) at approximately 434 MHz. Technology Tested Radio frequency (RF) technology was implemented for all of the proximity detection experiments. The system is made up of an in-cab device (EPU) for equipment and a PPU for ground workers. The EPU device contains a single antenna, reader, alert mechanism and can be connected to the central power source of a piece of equipment. The PPU device consists of a chip, battery, alarm and can be installed on the hard hat of a ground worker. A signal broadcasted by the EPU is intercepted by the PPU when the devices are in too close proximity, which is defined by the calibrated alert distance. The signal is broadcast by the EPU in a radial manner and loses strength as the distance increases from the EPU location. The proximity range can be manually modified by the user to lengthen or shorten the range at which an alert is activated. When the PPU intercepts the radio signal, it immediately returns a signal and an alert is activated from the EPU in real time. The proximity detection devices used for these experiments have two different alert methods. Forklift operators in a hazardous proximity location can receive a visual and audible alarm. Equipment operators receive an alert through an audible alarm and visual flashing lights located on the device inside the equipment cabin. The audible alert creates enough noise so that workers or operators wearing earplugs are still able to hear and distinguish the alert. The audible alarm is also different from other sounds and backup alerts common to construction jobsites. The visual alerts provide more alert options because the workers can become desensitized to audible alerts. A series of red lightemitting diodes (LEDs) activate upon a proximity breach. AIST.org October

5 These lights are distinguishable among typical equipment controls. This visual alert is shown in Figure 2 along with the PPU and EPU. The PPU and EPU devices were designed to be durable, including sturdy casing capable of withstanding daily weathering. The PPU rechargeable battery power duration is approximately two average workdays. A small LED located on the PPU is activated when the device is charged and working. The EPU can connect directly to the battery source of a piece of equipment and also displays a green LED when the system is functioning properly. The EPU unit can be installed in areas visible to the operator in the equipment cabin without obstructing the line of sight to objects outside of the cabin. The proximity detecting and warning technologies used in the experiments are capable of four different alert scenarios. These scenarios describe the action of the technology when the ground worker is located a safe distance from a piece of equipment and when the ground worker is in a hazardous proximity situation. The hazardous proximity region around a piece of equipment is pre-defined and calibrated into the proximity detecting and warning devices. The following four scenarios can occur when using these devices: True positive: An alert is activated when a ground worker is in too close proximity with a piece of equipment. False positive: An alert is activated when a ground worker is located at a safe distance from a piece of equipment (also referred to as nuisance alarms). True negative: An alert is not activated when a ground worker is located at a safety distance from a piece of equipment. False negative: An alert is not activated when a ground worker is in too close proximity with a piece of equipment. Of the four alert scenarios, false negative scenarios are the worst cases because the technology fails to alert the equipment operator of the hazardous proximity situation. False positives (nuisance alarms) are also undesirable because frequent non-hazardous alerts can desensitize equipment operators to potential hazardous proximity situations. Both true positive and true negative alerts are desirable alert scenarios because they accurately identify the status of proximity situations. Figure 3 presents a flowchart of the four possible alert scenarios for the designed experiments. Figure 2 PPU device installed on a ground worker s hard hat (left), and an EPU device installed on a piece of equipment (right) with alert activated. Figure 3 Safe distance between worker and operator Unsafe distance between worker and operator Flowchart of alert scenarios. Experiments and Results With Proximity Warning Devices A proximity detection device prototype was used based on the safety needs of the metal manufacturing industry. This system was evaluated in different experimental settings, each evaluating the performance and capabilities of different aspects of the system. Safe Proximity Zone Alert? Hazardous Proximity Zone Alert? Yes No Yes No False positive True negative True positive False negative 60 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

6 Figure 4 Test bed for field experiments. Field Condition Trials Stage 1: For the first experimental trials, an EPU device was placed in a simulated outdoor material handling environment to evaluate the effectiveness of the proximity detection system in the outdoor field conditions. The test bed for these trials was a clear, flat, asphalt paved surface with no obstructions. A robotic total station (RTS) and traffic control devices were used to create this test bed. The RTS was positioned at the center of a 15.2 m (50 foot) radius circle, and traffic control devices were placed at 36 equal distant locations around the circumference of the circle. The traffic control devices were positioned at 10 offsets around the circle. Figure 4 shows the test bed used for these experiments. The center point of the circular test bed served as the location for the EPU s antenna component. The EPU was installed inside the forklift cabin in view and audible range of the operator. Two antenna components of two different EPU devices were mounted on top the forklift at the highest point for the best detection range. These configurations can be viewed in Figure 5. The personal protection device was clipped to the hard hat of a worker on foot, as was done in the previous experiment. A worker wearing a hard hat equipped with a PPU approached the forklift at a constant walking pace from 36 approach angles of equal distance. After the EPU device Figure 5 Mounting positions of EPU antennas on a forklift. activated an alert, the worker stopped walking and measured the alert distance using a measuring wheel (Figure 6). This method was repeated three times for each approach angle. The total sample size for the forklift tests was 108 measurements. A statistical analysis was performed on the measured alert distance. The average value of the three trials was used to perform the statistical analysis. Results of the data analysis for the proximity detection alert distances of workers approaching a static forklift are: median = 9.7 m, range = 11.8 m and standard deviation = 3.2 m. The system recorded no false negative readings, meaning the alert was activated on every approach. The data obtained from these trials was used to develop proximity alert range graphs. These graphs display the recorded distant measurement from the worker s position to the EPU antenna at the time the alert is activated. Figure 7 shows the recorded alert distance measurements of the proximity detection and alert system. The EPU devices were installed on several other pieces of heavy equipment. The same test method as described above was used to evaluate the proximity detection and alert devices. Three different calibration ranges were tested on each piece of equipment. The calibration range with the highest coverage area percentage was chosen as the best calibration range of the three tested. If all three calibration ranges provided total coverage around the piece of heavy equipment, the range with the smallest standard deviation was selected. A statistical analysis for the selected calibration range for each piece of heavy equipment is shown in Table 3. The normal distribution curve was used to perform the analysis. False positive values were defined as the number of alert range distances greater than three times larger than the inner quartile range added to the upper quartile range value. False negative values were the number of times the test person contacted the heavy equipment before an alert was activated. Each calibration range has a sample size of 72 measurements. AIST.org October

7 Figure 6 Measurement of the alert distance. Figure 7 Forklift proximity warning zone for a specific calibrated alert range. 62 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

8 Table 3 Analysis of Proximity Alert Range for Various Pieces of Heavy Equipment Equipment Average Range Standard deviation False positives False negatives Forklift 9.74 m m 3.16 m 0 0 Wheel loader m m 3.98 m 1 0 Grader 8.67 m m 3.20 m 3 3 Asphalt paver m 7.51 m 1.70 m 0 1 Excavator m m m 3 5 Truck and trailer 7.25 m m 9.43 m 0 2 Pickup truck 8.43 m m 4.85 m 8 0 Vibratory roller m m 9.27 m 0 0 Mower m m 3.40 m 0 0 Dump truck m m 6.07 m 2 0 Field Conditions Trials Stage 2: The final sets of field experiments were conducted to test the effectiveness of the proximity detection devices on mobile equipment and static workers. These tests were completed on a flat, unobstructed, paved surface similar to the previous field experiment. The EPU device was installed in a pickup truck with the antenna mounted on top of the truck s cabin on the driver s side. A static ground worker equipped with a PPU was positioned next to an RTS, and was aligned on a straight path with the pickup truck. Traffic control devices were spaced 5 and 10 m along the truck s trajectory toward the ground worker. After maintaining a constant speed of 16 km per hour (10 miles per hour), the truck driver stopped the vehicle after the alert triggered. Three different proximity detection devices with varied alert ranges (short, medium and long) were tested in this experiment. Each of the three PPUs was tested 32 times. The left side of Figure 8 shows the stopped truck after the EPU has activated an alert. Box plots of the three different ranges and 96 data points gathered from this experiment are shown on the right side of Figure 8. All trials resulted in true positive alerts. No false alerts (including nuisance alarms) occurred during any of the 96 trial runs. Data Recording In addition to the location distance measured by the research team during all experimental trials, the proximity sensing and warning technology system has data recording and logging capabilities. During selected experiments previously described, the system recorded the ground PPU tag number, time and date of each proximity breach. A proximity breach included all instances in which a ground worker entered or exited the pre-calibrated hazardous proximity range around a piece of construction equipment. The column labeled Level in Figure 9 indicates the relative distance at which the system triggered an alert to the equipment operator. The column labeled status denotes if the worker entered ( Trip ) or exited ( Clear ) the pre-defined proximity distance of a piece of equipment. A sample of the data logged during these experiments can be viewed in Figure 9. Information from this system was exported into a database and further analyzed. Individuals involved with a large amount of proximity breaches when compared to co-workers were highlighted. Safety managers can use this data to further explore details of the proximity breach by monitoring the individual and providing further safety training specific to hazardous proximity issues to select Figure 8 Static worker and mobile equipment experiment (left), and box plot of proximity detection distance for each calibrated range (right). AIST.org October

9 Figure 9 distances for the field condition trials using the normal distribution. For each of these trials, the PPU were calibrated as follows: range 1: 640, range 2: 580 and range 3: 515. The software allows for calibrated relative alert distances between 0 and 1,000. Equipment Operator Visibility The workflow for this research component was divided into three sequential processes: (1) data collection, (2) data processing and (3) automated blind spot identification. This workflow measured blind spot data for heavy equipment, and transformed the raw data into operator visibility diagrams and quantifiable blind spot data. Software interface of recorded proximity breaches. ground workers and equipment operators. Safety personnel can also use this tool to provide preventive safety training by reviewing past recorded data and informing ground workers and equipment operators of past hazardous proximity situations. The proximity breach data record capabilities of the tested proximity detection and alert system are calibrated and function based on relative distances, or the distance from the PPU to the EPU relative to other detected devices. Proximity breach data for both stages of the field condition trials were recorded and analyzed. The relative distance at which the PPU device was detected can be viewed in Figure 9 under the column heading Level. An alert was triggered (trip condition) or silenced (clear condition) by the EPU at these relative detected distances. Results indicate that physical features of the equipment impact the correlation between the desired calibrated relative distance and the actual triggered alert relative distance. Table 4 shows an analysis of the calibrated alert distances and actual alert Table 4 Analysis of Recorded Relative Distances for Alerts of Field Condition Trials (without units) Range Average alert level Range of alert level Standard deviation of alert level Range Range Range Spatial Point Clouds: Data collection for blind spots used an automated blind spot measurement tool with a 3D laser scanner to evaluate different pieces of heavy equipment. The measurement procedure is illustrated in Figure 10. A commercially available 3D laser scanner recorded 3D spatial data from a position inside each piece of heavy equipment evaluated. The scanner s mirror was positioned at the operator s eye height as defined in the ISO standard. Previous experiments 14 found that the light source emitted by the laser scanners generally has no influence on the blind spot measurements. Depending on the type of the 3D laser scanner, it registers and stores, at a minimum, approximately 1.5 million data points per scan in a point cloud data file. These data values were registered and then converted to a standard X-, Y- and Z-coordinate system for blind spot analysis. The experimental design of collecting geometry information of the equipment followed the requirements from existing standards. The ISO code 5006 (Earth-moving machinery-operator field of vision test method and performance criteria), ISO code (Earth-moving machinery field of vision of surveillance and rear-view mirrors part 1: test methods) and ISO code (Powered industrial trucks test methods for verification of visibility 12 m radius) were used as standards for blind spot measurement during the experiments (ISO 2009). The ISO code 5006 provides a manual equipment blind spot measurement method that was developed through years of worldwide experience with construction equipment. This method uses two bright light bars spaced at an average human eye width inside the equipment cab. The light is projected onto a flat surface around the outside of the equipment cab to determine the visible areas to the equipment operator. Dark or shaded areas can be interpreted as blind spots for the equipment operator. The ISO code 5006 specifies a 12 m radius extended from the operator s position for the blind spot measurement area (ISO 2011). Each laser scan measurement took approximately 30 minutes (much faster than any previous method). An observed problem in preliminary experiments was to maintain the lifting fork height at the same elevation throughout the 64 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

10 Figure 10 Laser scan measurement of equipment cabin interior. experiment. Since the hydraulic lift forks release pressure over time, two small wooden (or metal) support plates underneath the bucket kept its position in place for the duration of the measurement. There were no other moving parts that could have influenced the point cloud measurement. An additional issue with laser scanners is its potential for limited field of view (FOV). Most laser scanners cannot measure the bottom parts of an equipment cabin, such as cabin floor or chair. Teizer et al. 14 have found a method to compensate for such pieces that belong to the cabin structure. All experimental tasks were executed using specifications and guidelines from both ISO codes 5006 and The experimental procedure differed in some aspects from the suggested ISO code 5006 manual measurement method. The 3D laser scanner emits a light source from one central location rather than using two light sources prescribed in the ISO code Previous research indicates that emitting a light source from one location better simulates stereo or singleness vision, which is more characteristic of human vision than two emitting light sources (Ogle 1950). Unlike the ISO code 5006 method, the 3D laser scanner also allows the user access to visibility data from any point inside the cabin. Although the manual measurement method was replaced with an automated blind spot measurement tool, the experimental strategy, measurement dimensions and results remained in compliance with the rules established by the ISO codes. Data Processing: The 3D point cloud blind spot data captured by the 3D laser scanner were processed through a previously developed automated blind spot detection and calculation tool. 14 This tool incorporated a ray-tracing algorithm to find areas not visible to the equipment operator. The algorithm traced an imaginary path of light projected from the driver s head position. All of the projected light rays are traced to detect if and where they intersected a previously measured laser scan point. If the projected ray intersected a laser scan point, the ray existing outside of the equipment cab was deemed a blind spot line. The volume outside of the equipment cab is divided into virtual cubes, creating a 3D grid. Because the laser point cloud penetrates the virtual three-dimensional grid, all of the point cloud data passes through one of the virtual cubes. Based on the visible or non-visible status of the point cloud passing through any given cube, the entire cube converts to the same visibility status. 14 The volume of each cube is projected as an area onto a 2D grid for analysis. These 2D grids are representations of the blind spot volume obtained by the laser scanner point cloud data. An average data processing time to evaluate the point cloud data and develop one blind spot maps is approximately 10 minutes for an experienced person with a modern computer (Dell T3500, 12 GB RAM, 1 GB visual graphics card). The 2D grids created for each skid steer loader can be viewed in a later section of this paper. Blind Spot Identification: The blind spots were identified based on the occupancy grid map achieved from the previous step. In addition, specific factors that were defined in ISO standards were computed to evaluate the design of the equipment cab. The terms and explanations of this section should be used to interpret the experimental results. Blind Spot Area(s): The invisible area(s) to the operator are shown by hash marks in the drawings. Rectangular Box (RB): A rectangular box is drawn around the equipment. The lines of the box are located on the ground reference plane at a 1 m distance from the outside rectangular boundary of the equipment. Visibility Test Circle (VTC): A circle with 12 m radius located on the ground reference plane with its center vertically below the center point of the operator s eye. Visibility Test Person (VTP): A test person with 1.5 m in height and 0.6 m in (shoulder) width. Visible Angles (Lengths) on the Circumference: The angles and lengths on the circumferences of the 12 m radius circle that is visible to the operator. Plan, Side and Front/Rear Views: The blind spot diagrams are displayed from different perspectives. The drawings provided in the results section are not to scale. Figures 11 and 12 demonstrate how to interpret the figures in the results section of this paper. The specifications listed on Figure 12 for the side view are the same for the front view figure. Forklift operators in steel manufacturing environments experience limited areas of visibility while performing work tasks. The infrastructure required to handle materials for steel manufacturing on forklifts impedes the line of sight for the operator. Figure 13 shows different views of inside a forklift typically found in material handling environments of steel manufacturing. Figure 14 shows the plan view of a forklift operator s visibility. An arrow is used in the figure to show the forward direction of the forklift (the direction the operator is facing). One side door remained open during the laser scan in order to install the laser scanner as prescribed by the ISO codes. Table 5 presents the measured forklift operator visibility. Limitations, Future Work and Application Areas The objective of this research was to evaluate the effectiveness of proximity detection and alert technologies on material handling equipment in the steel manufacturing environment. After completing the experiments, the research team found limitations to the tested proximity detection system. Many parameters and potential influences on the system should be evaluated through future experimentation. Future studies should include the following: AIST.org October

11 Figure 11 The plan view diagram. Figure 12 The side view diagram (similar for front/rear and left/right views). 66 Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

12 Figure 13 (a) (b) (c) (d) Operator views inside a forklift cabin: front view (a), rear view (b), side view (c) and top (d). Impact of temperature, humidity, precipitation and other ambient influences on warnings and alerts, and use of batteries on PPUs. Location and mounting positions of the EPU and PPU devices. Reaction of workers and equipment operators to implementing the devices. Calibration of specialized alert distances for individual pieces of equipment including operator and worker reaction time, operator brake distances, ground surface preparation, weather conditions and object mitigation strategy. Record and analyze near-miss data to further educate workers on proximity issues. Analysis of calibrated alert distance and actual alert distances on various pieces of equipment in different environmental settings. Collect and analyze nuisance alerts to evaluate reliability of system. Create an implementation strategy for proximity detection systems. Extended field trials with the proximity detection devices. Illnesses, injuries and fatalities resulting from hazardous proximity issues can become very expensive after summing medical costs, insurance costs, productivity decrease resulting from time lost and possible litigation costs. Some of these costs could potentially be avoided by implementing emerging safety technologies such as real-time proximity detection and alert systems. This safety technology can improve safety by giving equipment operators a warning during a hazardous proximity situation. Conclusion Current safety practices for proximity issues in steel manufacturing material handling areas have proved inadequate, as demonstrated by the number of fatalities, accidents and illnesses resulting from proximity issues. The ultimate goal of the steel manufacturing industry must be to achieve zero accidents and injuries. The purpose of this research was to determine a method that evaluates the capability of proximity warning and alert systems to function in the steel manufacturing environment and if they can provide alerts when resources (e.g., workers, equipment) are in too close proximity. Results obtained from the review and experiments suggest that the method of testing real-time proximity sensing and alert systems works successfully, and furthermore, that the tested technology has the potential to improve equipment-worker safety during material handling operations. AIST.org October

13 Figure 14 Plan view of forklift laser scan data. The proximity sensing and alert device prototype demonstrated its ability to perform by detecting the presence of hazardous proximity situations. In nearly all trials, the proximity detection system was able to detect and activate an alert when ground workers and material handling equipment were in too close proximity to each other. The audible alerts were to a sufficient volume to be heard over backup alarms and other general and loud manufacturing noise. The equipment operator was also able to see the visual alert provided by the EPU device. Overall, the system demonstrated its ability to perform in the simulated material handling environment by warning equipment operators Table 5 Summary of Plan View Forklift Operator Visibility Item Measurement Area of full circle m 2 Area occupied by equipment 23.8 m 2 Blind spot area m Length of circumference (full circle) 75.4 m Visible length circumference 49.6 m Percent visible circumference 65.8% Percent blind spot 60.0% when they were operating in too close proximity to ground workers. Although the field trials with the proximity detection and alert system were deemed successful, other parameters were noted that could potentially have an influence on the system, specifically the signal propagation. These factors include ambient temperature, relative humidity, mounting position and orientation of the devices on ground workers, equipment and other objects. These barriers along with others require further investigation to better evaluate the effectiveness of implementing proximity detection and alert devices in the steel manufacturing environment. References 1. U.S. Bureau of Labor Statistics, Injuries, Illnesses and Fatalities Current and Revised Data, Hinze, J., and Teizer, J., Visibility-Related Fatalities Related to Construction Equipment, Safety Science, Vol. 49, No. 5, 2011, pp International Organization for Standards, Earth-moving Machinery Field of Vision of Surveillance and Rear, International Standard, Part 1: Test Method, National Institute for Occupational Safety and Health, Highway Work Zone Safety, Center for Disease Control and Prevention, NIOSH, Recommendations for Evaluating & Implementing Proximity Warning Systems on Surface Mining Equipment, Report of Investigations RI 9672, Department of Health and Human Services: CDC, Atlanta, Ga., Iron & Steel Technology A Publication of the Association for Iron & Steel Technology

14 6. NIOSH, Preventing Injuries and Deaths of Workers Who Operate or Work Near Forklifts, National Institute for Occupational Safety and Health Publication No , Department of Health and Human Services, Cincinnati, Ohio, Occupational Safety and Health Administration, Materials Handling and Storage, Ogle, K.N., Research in Binocular Vision, Hafner Publishing Co., N.Y., Ruff, T.M., Monitoring Blind Spots A Major Concern for Haul Trucks, Engineering and Mining Journal, Vol. 202, No. 12, 2001, pp Ruff, T.M., Advances in Proximity Detection Technologies for Surface Mining, Proceeding of the 24th Annual Institute on Mining Health, Health, Safety and Research, Salt Lake City, Utah, Ruff, T.M., Evaluation of a Radar-Based Proximity Warning System for Off-Highway Dump Trucks, Accident Analysis & Prevention, Vol. 38, No. 1, 2005, pp Steele, J.; Derunner, C.; and Whitehorn, M., Stereo Images for Object Detection in Surface Mine Safety Applications, Western Mining Resource Center Tech Report No. TR , Colorado School of Mines, Golden, Colo., Teizer, J.; Allread, B.S.; Fullerton, C.E.; and Hinze, J., Autonomous Pro-Active Real-Time Construction Worker and Equipment Operator Proximity Safety Alert System, Automation in Construction, Vol. 19, No. 5, 2010, pp Teizer, J.; Allread, B.S.; and Mantripragada, U., Automating the Blind Spot Measurement of Construction Equipment, Automation in Construction, Vol. 19, No. 4, 2010, pp Teizer, J.; Caldas, C.H.; and Haas, C.T., Real-Time Three-Dimensional Occupancy Grid Modeling for the Detection and Tracking of Construction Resources, ASCE Journal of Construction Engineering and Management, Vol. 133, No. 11, 2007, pp Teizer, J.; Venugopal, M.; and Walia, A., Ultra Wideband for Automated Real-Time Three-Dimensional Location Sensing for Workforce, Equipment and Material Positioning and Tracking, Transportation Research Record: Journal of the Transportation Research Board, Vol. 208, No. 1, 2008, pp F Nominate this paper Did you find this article to be of significant relevance to the advancement of steel technology? If so, please consider nominating it for the AIST Hunt-Kelly Outstanding Paper Award at AIST.org/huntkelly. This paper was presented at AISTech 2012 The Iron & Steel Technology Conference and Exposition, Atlanta, Ga., and published in the Conference Proceedings. The Don B. Daily Memorial Fund Recipients Georgia Institute of Technology Dr. Jochen Teizer, Field Trials of Real-Time Proactive Warning and Alert Technology Lorain County Community College Duncan Estep, Customizing Occupational Safety Training for the Steel Industry University of Alabama at Birmingham Elizabeth H. Maples, Preventing Hand Injuries in the Steel Industry Through Assessment, Awareness and Training University of Washington Giovanni C. Migliaccio, Worker Health and Safety Monitoring in a Steel Manufacturing and Fabrication Environment Virginia Tech Alan P. Druschitz, Fall Prevention Systems for Flatbed Trailers, Phase 2 Implementation and Evaluation A no-cost, one-year extension supported by Steel Dynamics Inc. West Virginia University Daniel E. Della-Giustina, An Effective Safety and Health Program Through Total Quality Management to Assist Safety Professionals in Mitigating Risk of Injury to Upper Extremities (Hands and Arms) Within Steel Mill Operatons Don B. Daily Memorial Fund Committee The Don B. Daily Memorial Fund Committee, a joint committee of the Steel Manufacturers Association (SMA) and the AIST Foundation, administers the fund and selects the annual fund recipient(s). Richard P. Teets (chair), Steel Dynamics Inc. Butch Collins (AIST), Gallatin Steel Chad McClimans (AIST), AM Health and Safety Bradley Bray (AIST), California Steel Industries Inc. Kevin Burg (SMA), Charter Steel Shannon Johnson (SMA), SSAB Justin Hayes (SMA), Nucor Corp. Milwaukee School of Engineering Joseph C. Musto, Design of an Active Warning System for Fall Protection Sponsored by Charter Steel AIST.org October