State Estimation for Position Tracking

Independently developed a state estimation algorithm for tracking global position in a robot built in-house at Imperial College London Robot Intelligence Lab, found at: https://www.imperial.ac.uk/robot-intelligence/

​

The robot in use was known as SLIDER and featured novel sliding leg joints instead of knees. It is uniquely cheap to produce, which was a driving factor in its design. 

​

A visual SLAM algorithm known as ORB-SLAM2 (available at: https://github.com/appliedAI-Initiative/orb_slam_2_ros) was fused with IMU data in a Kalman filter to produce high frequency location estimates suitable for use in both balance control and path planning. It was implemented in C++ using ROS, enabling it's real-time operation. It was created in simulation due to COVID-19 however is planned to be tested on the real system later in 2020. The code can be viewed (but unfortunately not run) at: https://github.com/joeggg/FYP-State-Estimation-SLIDER