Human-in-the-Loop Shared Control with Guaranteed Safety for Teleoperated Robots

16/10/2025

Supervisor: Prof. Ville Kyrki (ville.kyrki@aalto.fi)

Advisor: Shivam Chaubey (shivam.chaubey@aalto.fi), Dr. Francesco Verdoja (francesco.verdoja@aalto.fi), Prof. Shankar Deka (shankar.deka@aalto.fi)

Keywords: Perception and Mapping, Uncertainty modelling, Optimization

Project Description

Shared control and human-in-the-loop systems are critical in high-stakes robotics, particularly for planetary rover teleoperation. These scenarios involve challenges like uncertain environments, strict safety constraints, teleoperation delays, and the possibility of unsafe user inputs.

In this project, we develop a shared control framework that guarantees safety using control-invariant sets (CISs) [1, 2], which are computed from the robot’s dynamics and an environmental model. The CISs ensure that unsafe human commands are overridden, while safe commands are executed normally.

The student’s main task is to develop the environment model needed to compute these CISs. This involves:

  • Using onboard sensors (e.g., LiDAR, RGB camera) to detect obstacles and free space,
  • Construct a geometric map of the environment by representing obstacles as a union of convex sets with the minimum possible number of components [3],
  • Model system disturbances and measurement noises for robust operation.

Although the long-term motivation is space robotics, the full pipeline will be implemented and tested on a differential-drive robot in indoor environments. We will evaluate performance under different environment configurations and simulated teleoperation delays to demonstrate safety, robustness, and real-time applicability.

Deliverables

  • Develop real-time mapping.
  • Develop a method to decompose non-convex sets into a union of the minimum number of convex sets.
  • Model system disturbances and measurement noise.
  • Conduct experiments on a differential drive robot.

Practical Information

  • Pre-requisites: Sensor fusion, Optimization, SLAM
  • Software: Python, MATLAB, ROS
  • Additional (Optional): Control Theory, Robot Dynamics

References
[1] T. Anevlavis and P. Tabuada, “A simple hierarchy for computing controlled invariant sets,” in Proc. 23rd Int. Conf. on Hybrid Systems: Computation and Control, ser. HSCC ’20, 2020.
[2] Chaubey, S., Verdoja, F., Deka, S., & Kyrki, V. (2025). MISC: Minimal intervention shared control with guaranteed safety under non-convex constraints (arXiv Preprint No. 2507.02438)
[3] Castagno J, Atkins E. Polylidar3D-Fast Polygon Extraction from 3D Data. Sensors (Basel). 2020 Aug 26;20(17):4819. doi: 10.3390/s20174819. PMID: 32858994; PMCID: PMC7506964.

Tags: