2021 Smoothing the trajectory of the service robot using artificial intelligence

Summary

This section is to be filled out by the project manager based on the summary document.

Company	Yanu
Project Manager	Project Manager
Project Team	Alvo Aabloo, Houman Masnavi, Karl Kruusamäe, Igor Rybalskii, Kirsi Zirel, Kristjan Laht
Challenge Tackled	Developing and testing robot hand trajectory optimization for ROS for smooth robot hand movement
Technology Used	ROS, MoveIt, TOPP-RA
Lessons Learned	Using and debugging general purpose planners is very difficult and requires a lot of specilized knowledge that is hard to get.
Result Published	We integrated optimal trajectory optimization with the ROS robot hand ecosystem
Target Group	Manufacturing floors, robot food providers
Diagrams/Photos
Video

Implementation Details

Description

• This is a ROS package that processes MoveIt trajectories into a trajectory that is optimal with respect to speed / acceleration / time limits. This is usually a nice to have, but for some robots like the Panda Robot, it's a necessary component for smooth operation. Panda robots require jerk-free trajectories but the using the default MoveIt trajectories fails with a lot of jerk. This is very problematic for service robots that need to carry things.

• This is built upon ROS and TOPP-RA

How to Install & Run

0. Create a working ROS workspace

1. Install toppra

   • Or create a toppra wrapper package by cloning toppra to catkin_ws/src/toppra-ros/toppra
   • and adding a CMake file with extra dependencies turned off to catkin_ws/src/toppra-ros

2. Clone this package into catkin_ws/src

   git clone https://github.com/ai-robotics-estonia/smoothing_the_trajectory_of_the_service_robot_using_artificial_intelligence

3. Setup MoveIt

4. in folder catkin_ws source ROS workspace source /opt/ros/noetic && source devel/setup.bash

5. build with catkin_make

6. run the ros service rosrun toppra_optimizer_ros toppra_optimizer_server

7. generate a plan using moveit

8. in another terminal run rostopic pub /optimize_trajectory/goal toppra_optimizer_ros/OptimizeActionGoal with the message contents being the result from step 7. (or use this example generated plan)

9. run the resulting trajectory on a real robot

Credits

If you use this library for your research, we encourage you to reference the accompanying paper A new approach to Time-Optimal Path Parameterization based on Reachability Analysis, IEEE Transactions on Robotics, vol. 34(3), pp. 645-659, 2018.

Licensing

MIT License

How to Contribute

Create issues / pull requests in this repo.

Testing

Provide code examples and how to run tests for your project.