U Toronto's sliding sensor trims continuum robot shape errors
A sliding sensor inside a soft robot lets you tune where it's most confident—cutting shape errors.
Sliding Sensors: Configurable Confidence in State Estimation for Continuum Robots, by Ella Walsh, Spencer Teetaert, Eric Diller, Timothy D. Barfoot, and Jessica Burgner-Kahrs, introduces mechanically reconfigurable sensing for continuum robots. The extended abstract notes that continuum robots often operate in uncertain environments, where accurate state estimation is essential for safe interactions—and estimation confidence is inherently spatially non-uniform. The authors present a concept hardware design showing that a sensor can translate longitudinally within a robot, allowing state-estimation confidence to be reshaped toward task-relevant locations. They demonstrate that varying the sensor location reconfigures confidence and that sliding the sensor back and forth reduces full-body shape estimation errors compared to a single fixed tip sensor. The work was accepted as an extended abstract at the 2026 IEEE 9th International Conference on Soft Robotics (RoboSoft).
- Sliding Sensors use a longitudinally translating sensor to reconfigure state-estimation confidence along a continuum robot
- Sliding motion cuts full-body shape estimation errors vs a single fixed tip sensor, per hardware tests at University of Toronto
- Accepted at IEEE RoboSoft 2026—a step toward task-aware uncertainty shaping for soft robotics
Why It Matters
Enables safer, more precise continuum robots in surgery and inspection by focusing sensing confidence where it matters most.