Building the Bulldog Bolt: Sensor Selection for an Autonomous Vehicle Balancing Power Requirements and Functionality

Electric vehicles and autonomous vehicles are two major innovations that are currently available to the general public or under development. While they can be separate products, it is also likely that many autonomous vehicles, if not all, will be electric vehicles as well. This is seen in the SAE/GM AutoDrive Challenge and its successor, the SAE/GM AutoDrive Challenge II; in both competitions, an electric vehicle, the Chevy Bolt, is provided to the collegiate teams, which they then work to turn into an autonomous vehicle. The combination of electric vehicles and autonomous vehicles creates certain challenges, among them the issue of powering sensors and the consequent impact on the vehicle’s electric powertrain, and in particular on the vehicle’s range. The various sensors required to provide data to the autonomous vehicle will deliver different types of information and draw varying amounts of power, and this needs to be carefully considered in the vehicle’s overall design, with sensors chosen to provide all needed information to design a vehicle that can operate safely, but also minimize the range reduction that would result from the sensors’ power requirements. This is addressed in the competition; in Year 1 of the AutoDrive Challenge II, one of the Mobility Innovation challenge tasks, the 0-0-0 Challenge, asked teams to analyze the sensor suite and its power draw. The work presented in this paper draws on that analysis conducted by Kettering University’s Bulldog Bolt team, leading to a sensor suite that is optimized for the scenario presented in the competition. The sensor suite is presented along with the specific reasoning leading to its selection, and a discussion of some of the tradeoffs that take place in that selection. The specific sensor suite presented in this paper would be of value for similar scenarios in which autonomous vehicles may be expected to operate, and the methodology presented could be useful in other scenarios to formulate appropriate sensor suites.