Autonomous Wheelchair
2D and 3D navigation of a robotic wheelchair
This project enhances autonomous wheelchair navigation and person detection using LiDAR-based perception and control algorithms. The system recieves a goal position from the BCI headset and then shall autonomously naviagte to that goal position.
2D and 3D naviagtion
2D and 3D navigation of a wheelchair robot in a cluttered environment. The video shows the naviagtion stack along with sensor readings.
Autonomous Navigation
- HectorSLAM for real-time mapping. This was chosen since it allows for map-creation without any explicit odometry.
- AMCL (Adaptive Monte Carlo Localization) for precise localization. see AMCL ROS
- PID Control for smooth and accurate movement of wheels on different terrains.
Person Detection
- Fine-tuned DR-SPAAM (attention-based autoregressive model) for detecting people in 2D LiDAR sequences.
- Reference: Person detection in 2d range data
System Impact
This system ensures safe and efficient navigation in dynamic environments.