# Learning to Detect Collisions for Continuum Manipulators without a Prior Model

## TL;DR

This paper presents a data-driven collision detection method for continuum manipulators that does not rely on prior models or environmental assumptions, using only sensory data from FBG sensors.

## Contribution

It introduces a novel machine learning approach for collision detection in CMs that eliminates the need for prior geometric models or environmental knowledge.

## Key findings

- Successfully detects collisions with unknown obstacles.
- Operates without prior geometric assumptions.
- Effective on a non-constant curvature CM with FBG sensors.

## Abstract

Due to their flexibility, dexterity, and compact size, Continuum Manipulators (CMs) can enhance minimally invasive interventions. In these procedures, the CM may be operated in proximity of sensitive organs; therefore, requiring accurate and appropriate feedback when colliding with their surroundings. Conventional CM collision detection algorithms rely on a combination of exact CM constrained kinematics model, geometrical assumptions such as constant curvature behavior, a priori knowledge of the environmental constraint geometry, and/or additional sensors to scan the environment or sense contacts. In this paper, we propose a data-driven machine learning approach using only the available sensory information, without requiring any prior geometrical assumptions, model of the CM or the surrounding environment. The proposed algorithm is implemented and evaluated on a non-constant curvature CM, equipped with Fiber Bragg Grating (FBG) optical sensors for shape sensing purposes. Results demonstrate successful detection of collisions in constrained environments with soft and hard obstacles with unknown stiffness and location.