Deep Movement Primitives: toward Breast Cancer Examination Robot

TL;DR

This paper introduces deep Movement Primitives, a novel learning-based approach for autonomous breast palpation robots, enabling effective and adaptable manipulation trajectories based on visual input, demonstrated through real-robot experiments.

Contribution

It proposes a new deep learning method for manipulation path planning that explicitly models trajectories from visual sensory data, improving upon existing techniques.

Findings

Outperforms state-of-the-art methods in reaching and palpating a breast phantom

Successfully generates manipulation trajectories from visual input

Demonstrates effectiveness through real-robot experiments

Abstract

Breast cancer is the most common type of cancer worldwide. A robotic system performing autonomous breast palpation can make a significant impact on the related health sector worldwide. However, robot programming for breast palpating with different geometries is very complex and unsolved. Robot learning from demonstrations (LfD) reduces the programming time and cost. However, the available LfD are lacking the modelling of the manipulation path/trajectory as an explicit function of the visual sensory information. This paper presents a novel approach to manipulation path/trajectory planning called deep Movement Primitives that successfully generates the movements of a manipulator to reach a breast phantom and perform the palpation. We show the effectiveness of our approach by a series of real-robot experiments of reaching and palpating a breast phantom. The experimental results indicate our approach outperforms the state-of-the-art method.