Pushbroom Stereo for High-Speed Navigation in Cluttered Environments

TL;DR

This paper introduces a high-speed stereo vision system for obstacle detection on small UAVs, capable of operating at 120 fps using a novel subset of block-matching stereo processing combined with IMU data to build full depth maps.

Contribution

The paper presents a novel stereo vision algorithm optimized for high-speed onboard processing, enabling real-time obstacle detection on small UAVs without external sensors.

Findings

Operates at 120 frames per second on a mobile CPU

Detects obstacles at a single depth efficiently

Builds full depth maps using IMU and state-estimation

Abstract

We present a novel stereo vision algorithm that is capable of obstacle detection on a mobile-CPU processor at 120 frames per second. Our system performs a subset of standard block-matching stereo processing, searching only for obstacles at a single depth. By using an onboard IMU and state-estimator, we can recover the position of obstacles at all other depths, building and updating a full depth-map at framerate. Here, we describe both the algorithm and our implementation on a high-speed, small UAV, flying at over 20 MPH (9 m/s) close to obstacles. The system requires no external sensing or computation and is, to the best of our knowledge, the first high-framerate stereo detection system running onboard a small UAV.