Dead Reckoning by Path Averaging in an Instrumented Small-Scale Test Vehicle
Open Access
- Author:
- Maransky, Stephen
- Area of Honors:
- Mechanical Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Sean N Brennan, Thesis Supervisor
Daniel Humberto Cortes Correales, Thesis Honors Advisor - Keywords:
- Dead Reckoning
Position Estimation
Wheel Radius Estimation
Small-Scale Test Vehicle
GPS
Intelligent Vehicles
Path Averaging - Abstract:
- The Global Positioning System (GPS) is a vital part of countless modern vehicle and robotic applications including consumer vehicles, farming, construction, and mining; but it is also highly unreliable due to signal blockage, multi-path errors, and other factors. This thesis presents an algorithm for estimating vehicle position through dead reckoning when GPS reception is temporarily lost on a path that has been traveled previously using estimates based on wheel odometry and an average path recorded by processing prior traversals. To test and demonstrate this algorithm, a small-scale test vehicle capable of highly accurate GPS position measurements and wheel odometry is developed. The vehicle is built using an off-the-shelf radio-control (RC) buggy as a platform, and it is retrofitted with high resolution optical encoders and a Real Time Kinematic (RTK)-enabled dual-band GPS receiver. The position estimation algorithm is split into three parts: wheel radius estimation, path averaging and dead reckoning along the average path. The wheel radius estimate typically converges to better than 1.5 mm (about 0.06 in) of accuracy within 12 seconds of moderate-speed straight-line travel. The position estimation algorithm can then use these radius estimates, the average path, and wheel odometry to estimate the vehicle’s position when GPS position is lost. The algorithm is tested in this work using data collected with the small-scale test vehicle. Results show the path-referenced position estimation to be effective in reducing dead reckoning errors along curvy paths, with significant improvement in performance – reductions in error by factors of 2 to 10 – for sharp curves and long durations of travel.