Instrumentation for Imitation Learning: Enhancing Training Datasets for Clothes Hanger Insertion
2026-05-22 • Robotics
Robotics
AI summaryⓘ
The authors study how adding sensors directly into objects can help robots learn tasks more efficiently. They focused on teaching a robot to insert clothes hangers, using 180 examples with and without these sensors. Their results show that robots using sensor data did better than those relying on just vision by 14-25%. Interestingly, the robot learned to use the sensor information on its own. They also found that adding data from sensor-aided experts can help vision-only robots improve and nearly match the sensor-aided ones.
robotic manipulationimitation learningdiffusion policiesinstrumentationteleoperationsensor integrationclothes hanger insertionmachine learning datasetsvision-based control
Authors
Remko Proesmans, Thomas Lips, Francis wyffels
Abstract
Large behaviour models have transformed the field of robotic manipulation, but prohibitive data requirements have thus far prevented a revolution similar to vision language models. We believe that instrumentation, i.e. sensor integration in objects, can provide invaluable state information and enable efficient learning for robotic manipulation. In this paper, we present instrumented imitation learning of clothes hanger insertion. Using 180 teleoperated demonstrations, we train diffusion policies with and without access to instrumentation data. Results show that policies leveraging instrumentation outperform vision-only counterparts by 14-25 %pt and exhibit greater task awareness. Crucially, a black-box imitation learning policy learns to prioritise instrumentation signals without explicit guidance. In addition, enhancing the teleoperation dataset with rollouts from an instrumented expert policy, enables a vision-only student policy to achieve performance comparable to the instrumented expert, thereby surpassing the original vision-only policy. These findings establish instrumentation as a promising strategy to enhance imitation learning for robotic manipulation. Datasets are available on Zenodo.