Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

   

2010 Research Summary

Planning and Control of Steerable Flexible Needles under Uncertainty and Partial Sensor Feedback in 3-D Environments with Obstacles

View Current Project Information

Jur van den Berg, Ajay Patel, Max Shimshak, Laura Greig, Pieter Abbeel and Ken Goldberg

National Institute of Health

This paper presents a technique for planning and controlling flexible steerable needles towards a target location in 3-D anatomy, while avoiding regions of sensitive tissue to reduce patient trauma. Our approach takes into account motion uncertainty (for instance due to tissue deformation, tissue inhomogeneity, needle torsion, etc.) and assumes that only partial and noisy sensor feedback is available. Our approach consists of three steps: First, we present a technique for optimal estimation of the 3-D position and orientation of the needle tip under motion- and sensing uncertainty. Second, we present a planner to compute 3-D needle paths that (i) maximally avoid sensitive tissues, (ii) arrive as close as possible to the target location, and (iii) are maximally observable given the partial sensor feedback model. Third, we derive an LQR feedback controller to optimally guide the needle along the planned path towards the target location. We demonstrate the performance of our approach in modeled prostate cancer treatment environments with simulated uncertainty.

Figure 1
Figure 1: Steerable needle paths reaching a target in 3-D anatomy avoiding sensitive and impenetrable tissue.