Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Active Chassis Control during stepping motions for a walking excavator
We have implemented a balancing algorithm for our walking excavator HEAP. However, we want to allow the driver to still use steps in difficult terrain. We want to detect shovel contact with the ground, and only balance on the two legs away from the shovel.
Keywords: control - walking excavator
Conventional walking excavators are very hard to control. The driver needs to control each leg joint individually. This requires lots of training and experience, and makes working with the tool while driving very slow.
We have implemented a chassis balancing algorithm, that keeps the chassis at a constant orientation and height, while minimizing contact forces with the ground. This increases work efficiency, and traction.
A video of the controller in use can be found at https://www.youtube.com/watch?v=5_Eq8CxKkvM&list=PLE-BQwvVGf8Gs-ThVoEFlNDQ9kVne5qNQ&index=5
In difficult terrain, keeping all legs on the ground might not be what the driver desires. Stepping motions with the shovel in ground contact are used, to get up steep hills or place the wheels in more desirable positions.
Currently, we detect whether the shovel is pushing on the ground and then turn off chassis balancing. For some tasks, it is desirable to still balance on the two loaded wheels (e.g. the hind wheels if the shovel is in contact near the front wheels).
Conventional walking excavators are very hard to control. The driver needs to control each leg joint individually. This requires lots of training and experience, and makes working with the tool while driving very slow.
We have implemented a chassis balancing algorithm, that keeps the chassis at a constant orientation and height, while minimizing contact forces with the ground. This increases work efficiency, and traction.
A video of the controller in use can be found at https://www.youtube.com/watch?v=5_Eq8CxKkvM&list=PLE-BQwvVGf8Gs-ThVoEFlNDQ9kVne5qNQ&index=5
In difficult terrain, keeping all legs on the ground might not be what the driver desires. Stepping motions with the shovel in ground contact are used, to get up steep hills or place the wheels in more desirable positions.
Currently, we detect whether the shovel is pushing on the ground and then turn off chassis balancing. For some tasks, it is desirable to still balance on the two loaded wheels (e.g. the hind wheels if the shovel is in contact near the front wheels).
- Detect shovel contact and location
- Implement controller balancing on two wheels and shovel (reusing work done on four-wheel balancing)
- Switch seamlessly between controllers
- Design intuitive joystick control concept allowing individual leg control, and roll angle
These tasks should first be tried in simulation, and later on the machine on Hönggerberg
- Detect shovel contact and location - Implement controller balancing on two wheels and shovel (reusing work done on four-wheel balancing) - Switch seamlessly between controllers - Design intuitive joystick control concept allowing individual leg control, and roll angle
These tasks should first be tried in simulation, and later on the machine on Hönggerberg
- highly motivated student
- experience in robotics/mechanics
- some programming experience (C++/ROS)
- highly motivated student - experience in robotics/mechanics - some programming experience (C++/ROS)
Please apply with your CV and transcript.
Simon Kerscher (simon.kerscher@mavt.ethz.ch)
Tom Lankhorst (tom.lankhorst@mavt.ethz.ch)
Please apply with your CV and transcript. Simon Kerscher (simon.kerscher@mavt.ethz.ch) Tom Lankhorst (tom.lankhorst@mavt.ethz.ch)