Key insights and resources taken from the IET Control and Automation Technical and Professional Network seminar in June 2015.
The automotive industry is moving up a gear; an increased use of electronics and software is enhancing a car’s safety, connectivity and environmental impact and these three mega trends are also enabling a growth in automated driving features.
Progress has seen a growth in on-road demonstrations, for example in March Delphi took part in a project that saw a vehicle drive 5,500km autonomously from San Francisco to New York.
“A fully automated vehicle is the ultimate robotics solution. As Technical Director of Automated Driving, I can see and feel the excitement in our organisation. Our teams in Europe, US and Asia are working together in this new domain, developing ideas and technology that increase safety, fuel economy and connectivity in the vehicle,” says Delphi’s Serge Lambermont.
As the industry moves towards higher vehicle autonomy the challenges for engineers and technologists currently focus on safety.
“The challenge for autonomous driving is meeting the high safety requirements needed for the wide-scale deployment of vehicles with high levels of autonomy,” explains Shad Laws, Innovation Project Manager at Renault Innovation Silicon Valley. “A part of our work uses envelope control methods to improve the ability of the vehicle to handle safety critical situations, using all of the vehicle’s manoeuvring capability to avoid collisions.”
“In the next few years I believe we’ll see active safety features become more advanced,” continues Lambermont. “Radar sensors will be able to track more objects and radar-vision fusion technology will provide the best performance in all-weather conditions. Vehicle to vehicle (v2v) and vehicle to infrastructure (v2x) communication will further increase safety and allow vehicles to communicate over radio, and see further ahead and around the corner.”
View Shad Laws' presentation: Towards AI Control: Integrated Path Planning and Following for Safety-Critical Performance
Developments are underway both in industry and academia. Michael Fisher, Director of the University of Liverpool’s Centre for Autonomous Systems Technology, notes that the crucial element of autonomous systems is that they make their own choices without human intervention.
“My work concerns the essential ‘autonomous’ aspect of autonomous vehicles/ systems, specifically the software at the core of such systems that now makes the decisions that human drivers used to make. How can we be sure that the system makes the right decisions? It is important to be sure what such a system decides to do, and that its activities are understandable and explainable. My work involves distinguishing the central autonomous decision-making software as an identifiable ‘agent’, and then carrying out formal verification to prove that the system will always make the required decisions.”
Changes are afoot and Michael believes that in the near future there will be a much greater realisation that autonomous systems are different from traditional systems and that they represent a step-change in technology.
“The key aspect of “autonomy” needs to be understood and the ability to carry out deep analysis of the autonomous core needs to be developed,” he notes.
“Not only is such analysis, for example through formal verification of the system’s core ‘agent’, essential to the assessment of safe decision making but will probably be important to both legal aspects, such as vehicle regulation, and ethical aspects, such as any life or death decisions that the vehicle must make.”
Shad adds that he believes we’ll move away from the idea that an autonomous vehicle is an oversized four-wheeled robot.
“As time goes on I think autonomous systems will increasingly exploit car-specific aspects in order to improve their performance and AI control is no exception,” he concludes.
View Serge Lambermont's presentation: Guidance and control for autonomous vehicles
View the rest of the presentations below by visiting www.theiet.org/autonomous-vehicles
Take a look at the IET Control and Automation Community and get involved with the latest discussions on Autonomous Vehicles.
Watch Axel Fuchs’ presentation on 3D LiDAR Sensing and Perception
Watch Professor Philippe Bonnifait’s presentation On the use of maps for perception and localisation
Read more resources and insights from the IET’s Transport Sector on Autonomous Vehicles
Watch the IET Control & Automation Technical Network’s Highlights film from their Autonomous Vehicles Seminar