An autonomous EV must be recharged for it to drive to a destination beyond the range of the battery system. To extend autonomous operation, an autonomous charging system was developed. A design requirement was that the system be made using consumer grade components that are common to the DIY IOT movement to decrease system cost. The design and manufacture of the autonomous charging system will be briefly discussed but is not the focus of this thesis.
The focus of this thesis is to investigate the relationship between the operating speed and the accuracy of the automation algorithm. Initial development focused on performance, but the algorithm runtime was more than ten minutes, which was too long. The only portions of the code that could be improved were the hunt cycles for the port cover and the port detent. During the hunt cycles, the algorithm uses feedback between a vision system and the kinematics of the robot. The feedback loop compares the BB centroid to the center of the camera’s FOV. The hunt is completed when the comparison drops below a defined threshold. An experiment was conducted using different combinations of high, medium, and low accuracy thresholds for each hunt. The hypothesis was that it was possible for the cycle time to be reduced by decreasing accuracy without sacrificing system performance.
Test results validated the hypothesis and the cycle time was reduced by 16% without impacting system performance. This was done by using the lowest accuracy parameter for the charging Port hunt and using the medium accuracy for the Detent hunt. During the experiment, additional improvements were identified for the software and the mechanical systems. The improvements were implemented prior to outdoor, full-cycle testing. Outdoor tests were then completed and verified that the implemented improvements along with the accuracy parameters that were the outputs from the test results decreased the full cycle time by 16%.