FURI | Spring 2025

Sensorless Estimation of Contact Wrench for a 3-DOF Robot Arm

Detecting the wrench (force and torque) between a robot arm’s end-effector and its environment is important for tasks like assembly and manipulation testing, but it is usually done with expensive force/torque sensors. This project aims to estimate these forces using only data from the robot’s joints (position, velocity, and control inputs) without external sensors. Two methods will be used: a momentum-based observer and an Extended Kalman Filter (EKF). These methods aim to provide accurate force estimation in real time without the need for expensive sensors. The effectiveness of each will be tested on actual hardware.