FOCUSING ON FULL AUTONOMY How to avoid these problems? An arguably better option is to skip Levels 2 and 3 altogether, and focus directly on Level 4 automation. This is a path that is being followed by other players in the field, such as Google, and nuTonomy, the company co-founded by the author, based on several years of work done Level 4 automation can be with colleagues and students at MIT AeroAstro. Instead of introduced incrementally, in relying on a human taking over when necessary, Level 4 â€œfull terms of complexity of operation autonomyâ€? automation requires the system to cope safely with scenarios, speed levels, fleet any situation it encounters. Human override may be allowed size, and other criteria. (e.g., during testing and development), but is not necessary for safe operations. While apparently harder than Level 2 or Level 3, developing Level 4 automation is a problem that is, in fact, easier to define, understand, and eventually solve. In a Level 4 automation system, the car would remain within the boundaries of an operational envelope it defines, and within which it knows it can operate safely. One may wonder whether such an approach would be too ambitious a step, and whether the incremental path through Levels 2 and 3 would be more reasonable. The answer is that even Level 4 automation can be introduced incrementally, in terms of, for example, complexity of the scenarios in which it operates, speed levels, fleet size, and other criteria. Many Level 4 autonomous vehicles are being deployed today, operating at low speed in closed or relatively controlled environments. Examples include the Ultra system shuttling passengers between terminals at Heathrow, the Navya ARMA shuttle on trial on the streets of Sion, Switzerland, and the Easymile EZ10 on trial in the Gardens by the Bay in Singapore and other locations. It is expected that first deployments of Level 4 systems will be in geographically defined locales (e.g., specific routes in a defined area of a city) for applications such as (local) mobility on demand, or first- and last-mile connectivity to transportation hubs. The technology to offer such services in a safe way is nearing maturity rapidly: nuTonomy is planning its initial commercial offering for 2018 in Singapore.
(Self)driving towards tomorrow
Published on Nov 18, 2016
Annual magazine review of MIT Aeronautics and Astronautics Department research and educational initiatives.