Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic Throwing with Robotic Material Handling Machines

Lennart Werner, Fang Nan, Pol Eyschen, Filippo A. Spinelli, Hongyi Yang, Marco Hutter

Abstract

Automation of hydraulic material handling machinery is currently limited to semi-static pick-and-place cycles. Dynamic throwing motions which utilize the passive joints, can greatly improve time efficiency as well as increase the dumping workspace. In this work, we use Reinforcement Learning (RL) to design dynamic controllers for material handlers with underactuated arms as commonly used in logistics. The controllers are tested both in simulation and in real-world experiments on a 12-ton test platform. The method is able to exploit the passive joints of the gripper to perform dynamic throwing motions. With the proposed controllers, the machine is able to throw individual objects to targets outside the static reachability zone with good accuracy for its practical applications. The work demonstrates the possibility of using RL to perform highly dynamic tasks with heavy machinery, suggesting a potential for improving the efficiency and precision of autonomous material handling tasks.

Dynamic Throwing with Robotic Material Handling Machines

Abstract

Automation of hydraulic material handling machinery is currently limited to semi-static pick-and-place cycles. Dynamic throwing motions which utilize the passive joints, can greatly improve time efficiency as well as increase the dumping workspace. In this work, we use Reinforcement Learning (RL) to design dynamic controllers for material handlers with underactuated arms as commonly used in logistics. The controllers are tested both in simulation and in real-world experiments on a 12-ton test platform. The method is able to exploit the passive joints of the gripper to perform dynamic throwing motions. With the proposed controllers, the machine is able to throw individual objects to targets outside the static reachability zone with good accuracy for its practical applications. The work demonstrates the possibility of using RL to perform highly dynamic tasks with heavy machinery, suggesting a potential for improving the efficiency and precision of autonomous material handling tasks.
Paper Structure (22 sections, 4 equations, 8 figures, 1 table)

This paper contains 22 sections, 4 equations, 8 figures, 1 table.

Table of Contents

  1. Introduction
  2. Contributions
  3. Related Work
  4. Robot Throwing
  5. Automation of Heavy Machinery
  6. Methodology
  7. System Description
  8. Simulation for Training
  9. Observations and Actions
  10. Reward and Terminations
  11. Training
  12. Sim-to-Real Transfer
  13. Actuation Delay
  14. Passive Joint Friction
  15. Experiments
  16. ...and 7 more sections

Figures (8)

  • Figure 2: Two-dimensional throwing controller deployed on the M545 excavator with overlayed payload trajectory.
  • Figure 3: Human operators using material handling machines to perform throwing in different applications.
  • Figure 4: The pipeline for training and deployment of the controller. The learned throwing controller outputs joint velocity commands. During training, a torque-based simulator is used with a joint velocity controller. During deployment, a low-level controller replaces the joint velocity controller in simulation to command the hydraulic valves.
  • Figure 5: The material handling test platform is a 12-ton multipurpose excavator with a material handling gripper. A wireless IMU module is attached to the gripper for state estimation of the passive joints. The DoF considered in this work are marked in the picture.
  • Figure 6: Identification of the passive joint friction in pitch / Y direction. Comparison of simulated oscillation and measurements from the IMU.
  • ...and 3 more figures