The Smart Wheelchair Project

System Overview

In this project, we are exploring the use of synthesized landmark maps for localization of a smart wheelchair in outdoor and urban environments. Three dimensional maps are generated using data acquired from high fidelity sensors and GPS. The resulting maps are then synthesized to extract salient features, such as poles and building corners, for use as landmarks in localization. By leveraging such detailed maps and landmark data a smart wheelchair can localize itself inside the map with a much a less expensive sensor suite. This design allows us to gain many of the benefits of higher fidelity sensors and reduce the overall cost of the system.

3D Range Map Generation

Data was acquired for 3D map generation using Little Ben(below left), Ben Franklin Racing Team's entry into the DARPA Urban Challenge. Little Ben was equipped with a Oxford Technical Solutions RT-3050 GPS/IMU that provided accurate pose and GPS position data. It was also equipped with two Sick LMS291-S14 LIDAR systems mounted vertically to provide range and remission data. The data from the front (red) and rear (blue) LIDARs were registered with the pose data to produce the maps seen below(right).

Landmark Synthesis

We employed a recursive clustering approach in conjunction with several validation gates on cluster size and vertical level to segment pole-like features. The below figure shows the synthesized map data. In this experiment, no false positive or negative segmentations occurred. The relative distance error between synthesized landmark pairs and their real-world counterparts was about 7 cm. .

Localization Approach and Experimental Results

We used an Extended Kalman Filter (EKF) approach for estimating the 2-D wheelchair pose. In the prediction step, linear and angular velocities were estimated from the encoders. For the correction step, the observation functions were based upon LIDAR estimates for the range and bearing to the segmented landmark at a given position. Localization results using refined landmark position estimates shown below. Compared to measured ground truth, the wheelchair was able to estimate its global position with a mean error of 20 cm in the complete absence of GPS.

Previous Work: ATRS and DARPA Urban Challenge

This project is based on the previous work done as part of the ATRS and DARPA Urban Challenge projects.

Videos

Publications

Media Coverage

Sponsors

This work was supported in part by National Science Foundation CAREER Award #0844585. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

People