Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
A Two State Information Filter for Legged Robots
The aim of this project is to implement, test and verify a two-state information filter (TSIF) for a quadruped robot and a walking excavator to estimate the pose and velocities of the robots, as well as model parameters such as inertias and location of center of mass.
Keywords: State Estimation, Filtering
The goal of this project is to improve the current state estimation for the quadruped robot ANYmal and the walking excavator M545. The task is to replace the Extended Kalman Filter with a two-state information filter and to extend the state of the filter to co-estimate model parameters, which are important for model-based control. In particular, the mass, the inertia, and the location of the center of mass of the individual bodies are interesting to estimate because these parameters are difficult to measure and can change over time. For instance, the mass of the boom of the excavator changes when it is digging and transporting soil in its bucket.
The goal of this project is to improve the current state estimation for the quadruped robot ANYmal and the walking excavator M545. The task is to replace the Extended Kalman Filter with a two-state information filter and to extend the state of the filter to co-estimate model parameters, which are important for model-based control. In particular, the mass, the inertia, and the location of the center of mass of the individual bodies are interesting to estimate because these parameters are difficult to measure and can change over time. For instance, the mass of the boom of the excavator changes when it is digging and transporting soil in its bucket.
- Implement and verify a TSIF for ANYmal, which fuses IMU measurements with kinematic data.
- Implement a time extrapolation of the state.
- Include the dynamics in the TSIF to co-estimate the inertia and location of center of mass of the main body of the robot.
- Test and validate the filter on ANYmal together with the locomotion controller in the loop.
- Include external 6D pose measurement from ICP-based localization or GPS in the filter.
- Include visual information in the filter to reduce drift.
- Find and evaluate new methods for contact estimation.
- Optionally, test the filter on the walking excavator.
- Implement and verify a TSIF for ANYmal, which fuses IMU measurements with kinematic data. - Implement a time extrapolation of the state. - Include the dynamics in the TSIF to co-estimate the inertia and location of center of mass of the main body of the robot. - Test and validate the filter on ANYmal together with the locomotion controller in the loop. - Include external 6D pose measurement from ICP-based localization or GPS in the filter. - Include visual information in the filter to reduce drift. - Find and evaluate new methods for contact estimation. - Optionally, test the filter on the walking excavator.
- Good C++ programming skills
- Basic knowledge about filtering and optimization
- Good C++ programming skills - Basic knowledge about filtering and optimization