Mobile robots are increasingly being deployed in environments hazardous to humans. However, many of these robots require remote control operation or are tethered, requiring the human operator to remain within a potentially hazardous radius of the area of operation. To resolve this issue an Autonomous Robot Retrieval System (ARRS) utilising Open RatSLAM based on the Lego NXT 2.0 robotics platform is proposed but could not be implemented due to memory limitations of the hardware. An occupancy grid implementation was explored but conflicts between competing tasks prevented its successful operation. In addition, the Queensland University of Technology MatLab RatSLAM implementation, the precursor to Open RatSLAM, was investigated. Updates were made to the implementation to enable the reading of video files using updated functions, and its performance with regard to accuracy of mapping and running-time was assessed. The accuracy of mapping was found to vary greatly dependent on the nature of the video footage used and running-time of the program was found to be 10 times that of the video file duration. Both of these factors suggest that the MatLab RatSLAM implementation is not suitable for real-time SLAM operations.
Autonomous Robot Retrieval System
2015-11-10
Presented at: 2015 SAI Intelligent Systems Conference (IntelliSys 2015). (2015)
Bild
Elektronische Ressource
Englisch
DDC: | 629 |