The SmartWheeler platform. Collaboration between McGill, U.Montreal , Ecole Polytechnique Montreal + 2 clinical rehab centers. Standard commercial power wheelchair with onboard computer and custom-made electronics. - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The SmartWheeler platform• Collaboration between McGill, U.Montreal, Ecole
Polytechnique Montreal + 2 clinical rehab centers.
• Standard commercial power wheelchair with onboard computer and custom-made electronics.
• Sensors: laser range-finders, sonars, RGB-D camera (Kinect), wheel odometers.
Two primary components of cognitive robotic system:Interaction Manager and Navigation Manager
Current system:classical mapping path planning
Obstacle detection:Fusion of laser, sonar and RBG-D data to detect standard obstacles, including glass walls.Next challenge is to detect “negative” obstacles (holes, descending ramp)
Mapping and localization:Particle filtering SLAM algorithms, using laser data, integrated in ROS.
Local control:Pre-programmed behaviors for simple tasks (wall following, door traversal)
Global path planning:Deterministic search algorithm to achieve point-to-point navigation in discretized state space.
Ongoing research:Socially adaptive navigation
Approach:
- Extract information about dynamic obstacles from RBG-D camera.
- Acquire training data containing trajectories of driving in crowds from a human expert.
- Apply machine learning methods on the training data to learn a policy that matches dynamic features to control actions.
[Kim and Pineau., RSS 2013.]
Speech-based user interaction
Using a Partially Observable Markov Decision Process (POMDP) to model the dialogue interaction and select the robot response (right column).