iSeat - Towards an intelligent driver seat for autonomous cars

According to the most optimistic predictions, the first commercially available fully-autonomous cars are expected in 2040 offering the consumer a full end-to-end journey. These self-driving vehicles will be equipped with technology allowing autonomy Level 5 in which there is no interference required by the human. The concept of autonomy levels was first published by the international Society of Automotive Engineers in 2014. The report defines six levels of autonomy that automakers would need to achieve on their way to building the no-steering-wheel self-driving bubble pods of the future reaching from the fully-manual Level 0 to the fully-autonomous Level 5.

In the race towards the first commercially available fully-autonomous car, the majority of cars on UK roads will be equipped with technology that allows Level 3 or 4 autonomy over the next two decades. Drivers will be provided with increasingly sophisticated features such as lane-keep and steering assist. These semi-autonomous cars might be able to transport the driver autonomously on sections of a journey. However, the driver is required to take control occasionally between different levels of autonomy when required to complete an end-to-end journey. These transitions between autonomy levels cause safety concerns, as the driver might not be fully aware of the surrounding situation and the enabled autonomy features instantly.

This project proposes a new interface design for semi-autonomous cars called iSeat. This system is fundamentally different compared to current systems (such as Tesla's Autopilot or DistronicPlus by Mercedes) using visual or auditory indications which might be mentally overloading and distracting for the driver. iSeat is an intelligent driver seat acting as a co-pilot measuring the current mental and physical engagement of the driver and allowing safe, coordinated and timely transitions between different levels of autonomy. Of particular significance is the driver seat made of robotic structures serving the feedback purpose as well as providing monitoring capabilities through direct contact with the human during any level of autonomy: Tactile sensation can be fed back to the driver, the seat ergonomics and comfort can be changed and the robotic structures can measure the pressure distribution of the driver's weight. iSeat sensing information will be fused with multi-modal sensing data from electrical activity produced by skeletal muscles (Electromyography (EMG) signals) and in the driver's brain (Electroencephalography (EEG) signals), and input from vision cameras regarding the driver's posture and the point of gaze (i.e. where the driver is looking). This real-time knowledge will be classified through machine learning and affective interaction techniques in terms of the awareness state of the driver. Personalised feedback will be provided (i.e. tactile sensation, stiffness feedback, change of the driver seat ergonomics/comfort, visual/auditory feedback) to support the driver so that safe, timely, effective and intuitive transitions between different autonomy levels can be completed.

The iSeat system builds upon a complete re-think of the manner in which humans interact with autonomous cars. The smart combination of sensor systems, machine learning, affective computing, human factors, haptics and robotics will result in a bi-directional human-machine cooperation that is safe, intuitive, effective, and personalised.

Who will benefit from the project?

This outcomes of this project will benefit (i) the general public, i.e. people who will travel in autonomous cars that cannot offer full end-to-end journey, (ii) the UK industry, i.e. carmakers, aiming at producing autonomous cars, and the technology industry, that supplies carmakers with systems that will be integrated into cars and (iii) academics in the area of human factors, robotics and transport engineering. Other major beneficiaries will be policy makers who will investigate ways to standardise procedures in autonomous cars to ensure safety.

How will they benefit from the project?

The general public - Autonomous vehicles will transform the transportation industry beyond the automotive and technology industries and affect delivery services and public transport even in remote areas and dispersed communities. An autonomous vehicle will certainly be, for the majority of people, the first "autonomous robot" that they are interacting with on a daily basis and that might make decisions through AI that can be life-threatening. To overcome any lack of trust and confidence in these autonomous robots, which might lead to any constraints of public acceptance, iSeat addresses human-machine cooperation risk factors and introduces intelligent sensing and intuitive (haptic) feedback technologies into the driving environment.

The UK industry - The annual economic benefits of autonomous cars is predicted to grow to £51 and £121 billion by 2030 and 2040 respectively. Reliability and safety concerns can be extremely damaging for carmakers. For instance, after Tesla's Autopilot crash in 2016, new tighter regulations were released on testing automation features. Tighter regulation could slow the development and introduction of new technology. In worse cases, new developments might be aborted due to high-risk commercialisation factors. Further, autonomous cars are transformative for the automotive industry at large - acting as a catalyst for new products and technology. The introduction of autonomous cars will revolutionise safety. iSeat aims at reducing safety risks involved in the process of taking back control from a car or vice versa, which will lead to direct economic impacts to various road industries such as the automotive, coach, trucking, ground shipping, as well as, other transportation sectors such as aviation, maritime and rail. This will help the UK to become world-leading in the development of technological products and to attract experts for human-machine cooperation of autonomous cars.

Academics - Due to the high-risk nature of the proposed work, this project will primarily produce new knowledge and understanding on how various feedback modalities can support drivers to gain the right level of engagement/vigilance, optimally complete procedural steps between different levels of autonomy, and efficiently get back into the control loop in various driving situations. The knowledge gained throughout this project will be beneficial for other application areas such as the integration of collaborative robots that work closely together with the human in industrial settings.