Autonomous vehicles (AVs) will improve the safety of humans and help protect the environment, reducing crashes, congestion, and emissions. But autonomous systems can swim or fly as well as drive.
Behavioral planning is one of the biggest issues that Michigan researchers deal with in developing such vehicles. How does an AV or a robot decide what to do and how to do it? Robots don’t understand the intentions of the people around them. Neither do they understand how to behave in a way that helps the people around them understand the robots’ intentions. Michigan attacks this challenge by developing safe systems that can perceive and track the actions of the people around them.
This is a new area of robotics. Traditionally, roboticists determined what they wanted their machines to do, and then computed the lowest-energy way to do it. The presence of people in the robot environment hadn’t entered the equation in a major way. Nevertheless, people have their own goals, interests, and levels of safety and comfort when encountering robots that the latter need to account for.
A decade ago, Michigan roboticists also developed what they believed to be the first autonomous aircraft that could initiate and perform its own take-offs and landings on water. Called the Flying Fish, two prototypes of this seaplane were successfully developed to perform robotic surveillance over the ocean.
AV research at U-M includes other types of aviation-related projects and unmanned aircraft. Autonomous systems today can be designed to have logic and decision-making capabilities for handling contingencies and emergency flight planning for aircraft. Such capabilities may help reduce airplane crashes that are caused by pilot error, either as a sole factor or, as happens more often, as part of a combination of factors.
U-M engineers see a potential role for machine learning in aviation to help distressed drones find flat rooftops in urban areas to make safe landings. This work prepares for the day when many drones will fly low over cities, delivering packages and performing surveillance, while autonomous helicopters ferry passengers from downtown areas to major airports.
Robotics advances at Michigan led to the founding of May Mobility, which offers autonomous shuttle systems designed to alleviate traffic congestion in cities that are underserved by buses, light rails and scooters. In Detroit in 2018, May Mobility became the first AV company to launch a commercial fleet of shuttles on public streets in mixed traffic. In Columbus, Ohio, in 2019, May Mobility also became the first nation’s first AV company to allow rides to anyone without signing a liability waiver. Yet another first: the company provides AV service that’s integrated with a mass transit system, at the Amtrak station in Providence, Rhode Island.