11/12/13 15:11 Cloud Based Localization for Mobile Robot in Outdoors Xiaorui Zhu (presenter), Chunxin Qiu, Yulong Tao, Qi Jin Harbin Institute of Technology Shenzhen Graduate School, China
11/12/13 15:11 Outline 1 Background 2 Cloud-based Localization 3 Experiments 4 Conclusion
Background 11/12/13 15:11 Localization is an important problem for outdoor mobile robot 1.Traditional localization techniques: GPS Advantage: no accumulative error Disadvantage: signals unavailable in some cases Laser scanner Advantage: high precision Disadvantage: time-consuming Vision system Advantage: ample information Disadvantage: dependent on the illumination ; time-consuming
11/12/13 15:11 2. Long-term autonomous: Challenges: Adapt to the dynamic scenarios Detect and track static or moving objects (Zhao et al., 2008 ) Learn more knowledge to predict the environmental changes (Sunderhauf et al., 2012) Run a long time Require a large storage space Increase computational payloads 3. Cloud robotics: Offload computation: SLAM: feature extraction and filter for the state estimation (Arumugam et al., 2010) Access to large databases: Recognize and grasp objects: preprocess the object and access the large databases (Kehoe et al., 2013) Share the knowledge : Multi-robot negotiation: The well-equipped robot can get more ample information than the poor-equipped robots (Wang et al. 2012)
Background Main contributions of this paper: " Propose a new cloud-based architecture to achieve longterm autonomous localization potentially. " Propose a new way to obtain the latest map information.
11/12/13 15:11 Outline 1 Background 2 Cloud-based Localization 3 Experiments 4 Conclusion
Cloud-based Localization Architecture Fig 1. Cloud-based localization Architecture
Cloud-based Localization Algorithms 1. Offline phases: (1) Label a set of points along the roads in the selected area; (2) Extract the geodetic coordinate of these points from the Google Earth; (3) Update the road network map in the cloud. Fig 2. Road networks previously existing Fig 3. Updated road networks from the Google Earth
11/12/13 15:11 Cloud-based Localization Architecture 2. Online phases (1) Send the initial position estimated by the GPS to the cloud; (2) Pull the estimated initial position to the nearest road point as the initial position of robot (in the cloud); (3) Extract a local road network map within a certain distance around the initial position of the robot (in the cloud); Fig 4. Extracted a local road network map
(4) Compute the Robot-Terrain Inclination (RTI) model (in the cloud); Fig 5. Robot path is segmented into a series of line segments Fig 6. Geometric extraction of robot -terrain inclination model
(5) Send the RTI model to the robot; (6) Achieve particle filter localization based on the RTI model (on the robot). A. Motion Model: B. Sensor Model:
Outline 1 Background 2 Cloud-based Localization 3 Experiments 4 Conclusion
Experimental Setups Platform: Scenario: Mobile robot(summit XL); IMU(NAV 440) Travel distance: 500 m Robot speed: 1.0 m/s Fig 7. Experiment platform Fig 8. Google Earth and the point sets on the pre-planned path
Experimental Results Fig 9. The estimation of the robot position by the proposed technique
Experimental Results 11/12/13 15:11 Fig. 10 The position estimation errors of the robot using the proposed technique
Outline 1 Background 2 Cloud-based Localization 3 Experiments 4 Conclusions
" Conclusions " This paper introduces a cloud-based outsourcing localization technique for a mobile robot on outdoor road networks. " Experimental results validate the proposed technique and illustrate that the proposed technique has capability to achieve online localization. " Future works " This method will be applied to more complex large-scale/long-term circumstances.
1. H. Zhao et al., "SLAM in a Dynamic Large Outdoor Environment using a Laser Scanner," presented atinternational Conference on Robotics and Automation (ICRA), 19-23 May. 2008, Pasadena,USA, 2008. 2. N. Sunderhauf, P. Neubert, P. Protzel, "Predicting the change-a Step Towards Life-Long Operation in Everyday Environments," presented at 2013Robotics: Science and Systems (RSS) Robotics Challenges and Vision Workshop, 14-18 May. 2012, Minnesota,USA, 2012. 3. R. Arumugam et al., "DAvinCi: A Cloud Computing Framework for Service Robots," presented at International Conference on Robotics and Automation (ICRA), 3-8 May, Anchorage,USA, 2010. 4. B. Kehoe et al., "Cloud-Based Robot Grasping with the Google Object Recognition Engine," presented at International Conference on Robotics and Automation (ICRA), 6-10 May, 5. L.Wang, M. Liu and M. Meng, "Towards Cloud Robotic System: A Case Study of Online Co-localization for Fair resource Competence," presented at International Conference on Robotics and Biomimetics, 11-14 December, Guangzhou, China, 2012.Karlsruhe, Germany, 2013.
Thank you! 11/12/13 15:11