Industry · Robotics & Control Engineer · 2026–present
Autonomous smart wheelchair
Navigation and motion control for an electric wheelchair at a robotics startup.
The problem
A powered wheelchair that drives itself has to be smooth before it is fast: it carries a person, the load changes from one passenger to the next, and it moves among people and furniture.
What I did
- Navigation stack built on ROS 2 and Nav2
- LiDAR localization and global path planning
- Sensor fusion of wheel odometry and IMU for state estimation
- MPPI controller tuning for tracking, smoothness and obstacle avoidance
- Simulation scenarios for replanning, costmaps and behavior-tree recovery
- Dynamics model identified from data recorded on the prototype
Results
The navigation pipeline runs end to end in simulation, and the identified dynamics model keeps the simulated chair close to the real one. This is ongoing industry work, so details stay private.