Kinodynamic Trajectory Optimization and Control for Car-Like Robots

This video presents kinodynamic motion planning and predictive control based on Timed-Elastic-Bands.
In particular, the approach is applicated to a Lego Mindstorms Ev3 car-like robot.
Significant parameters of the robot are:
- width: 0.2 m
- length: 0.28 m
- maximum steering angle: 0.6 rad
- wheelbase L: 0.19
- maximum velocity: 0.15 m/s
The robot is localized with an OptiTrack motion capture system.
Nine boxes are representing static obstacles which are also tracked by the motion capture system. Notice, only obstacles within the perceptual range of the robot are considered for planning (see red cells in the visualization frame).
Each time a new goal is commanded to the system, the global planner plans a coarse path to the goal without taking the kinematics of the robot into account. The local planner (TEB) obtains intermediate goals from the global planner within a perceptual range of 1 m.
Transitions with different start and goal positions are investigated.
The approach is presented in
[1] Rösmann, C., F. Hoffmann and T. Bertram: Kinodynamic Trajectory Optimization and Control for Car-Like Robots, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, Canada, Sept. 2017.
Refer to ROS Wiki for more details on the software: wiki.ros.org/te...
The Institute of Control Theory and Systems Engineering (RST) at TU Dortmund is engaged in fundamental research in the fields of robotics and computational intelligence. Furthermore, the applied research focus is automotive systems and mechatronics. www.rst.e-techn...

[@solix777]

could you share your config file for this set up?

[@gokulnarayanan2844]

I had a note that you had 10 cameras for Localization, instead why can't you use a single 360 LiDAR, which could improve the computation speed rather compared to 10 cameras. Do u get any control loop frequency miss error by any chance appart from the feasible distance error. Please let me know, I had the same issue while testing, but after tuning a little bit I could improve them.

[@evilby]

I guess that they wanna apply the method to real cars, in which using cameras is a much economic solution.

[@soul0195]

What software have u used to path tracking

[@gokulnarayanan2844]

teb_local_planner I suppose