The fully autonomous vehicle (AV) is coming, and early models will be on roads sometime around 2020. To reach this milestone, automakers are turning to Silicon Valley for its expertise in connected devices, the Internet of Things, and human-machine interfaces. A recent tour of the 18-month-old Nissan Research Center (NRC-SV) in Sunnyvale, California underscored the importance of this trend in relation to the automotive industry’s development of the AV of the future.
While some autonomous drive systems that rely on cameras, lasers, and sensors, such as lane keeping and automatic braking, don’t require vehicle connectivity, the fully autonomous vehicle will.
The fact that AVs are likely to be far safer than non-autonomous, human-driven vehicles has been well-established. However, to provide the type of near guaranteed safety the auto industry and customers require, the fully autonomous vehicle described by NRC-SV Director Martin Sierhuis will be “the most complex system in the world.”
For starters, the fully autonomous vehicle needs to be able to communicate with other vehicles and infrastructure, anticipate/predict human and non-human (animal) behavior, be personable, constantly observe and relay information back to the Internet, and act quickly upon information received from all these sources.
Watch for Deer
Information on weather conditions, traffic congestion, and road construction are valuable assets to other vehicles and, in an ideal system, can be transferred seamlessly. Further, observations made by vehicles can be used to maintain a near real-time map of the world, given changes to road infrastructure. However, the most valuable pieces of information will be on how AV predictive systems can be improved and how AVs fail.
A major challenge for AVs is the unpredictability of the world. The awkward four-way stops, the sudden trajectory deviations, the deer on the side of the road, the ear buds-wearing bicyclists in downtown San Francisco, etc. all have to be accounted for. To function effectively, an AV must be able to predict both human and animal behavior better than humans do. Predictions are based on data; as more data is accumulated on humans through AVs, they will in turn be better able to predict human behavior and, therefore, safer in the more pedestrian-centric urban environments.
The above are all examples in which the sharing of information from AV to AV will avoid catastrophe; however, it must be assumed that failures will eventually happen. Yet, the silver lining will be that when the AV eventually does fail, the circumstances of that failure will be shared, and the overall system will learn from it. As Sierhuis explained, “The same accident will never occur again.”
Tags: Autonomous Vehicles, Clean Transportation, Electric Vehicles, Smart Transportation Program
| No Comments »