Handover process the key to connected autonomous vehicle development

‘Handover’ is when the driver of an autonomous vehicle takes back control. At speed this is far from straight-forward latest trials indicate.

The VENTURER team undertaking tests of AV
The VENTURER team undertaking tests of AV 'handovers'
  • Updated: 23 August 2017
  • Author: Peter Blackley, senior transport planner, Atkins' Transport division

Only two years ago government announced the first wave of funding to test connected and autonomous vehicles (CAVs) in the UK. £20m was split across three projects with VENTURER, an Atkins’ led consortium, being one of the first UK projects to explore the feasibility of driverless cars.

There has since been an unprecedented interest in this emerging market with the establishment of the Centre for Connected and Autonomous Vehicles (CCAV) and a steady stream of government funding to support R&D. Today, there are over 30 projects co-funded by UK government undertaking feasibility studies, trials and testing of CAVs.

This pattern is repeated across the globe with countries such as the US, Dubai, Singapore and China investing heavily in this market.

For many people there is often an assumption that when driverless cars are on the network all vehicles will be fully autonomous. This is known as Level 5 (SAE).

However, for those working in the industry there is still a level of uncertainty around reaching Level 5 and how to address some of the key issues surrounding Levels 3 and 4 of autonomy. At these levels, transfer of control between the vehicle and the human driver –known as the handover process – would be required as the vehicle isn’t fully autonomous.

Understanding the handover process is important from a safety, traffic management, technology and legal and insurance perspective. For example, the length of time it takes someone to regain full control of the vehicle represents a meaningful risk to insurers. Importantly, understanding when control is transferred between the vehicle and driver has liability implications.

To date, research on handover has focused on more experienced drivers at high speeds and involving single handover requests. This is not necessarily typical of the day-to-day driving experience in urban areas.

Therefore, VENTURER decided to undertake its first set of experiments with a focus on:

  • Drivers with varying levels of experience
  • Lower speeds (20, 30, 40 and 50 mph) typical of an urban environment;
  • Driving simulator and road experiments
  • Shorter driving periods with multiple handover requests

This first trial sought to gain a deeper understanding of:

  • How long it took participants to engage with the driving controls (steering wheel, brake, and accelerator) after a handover request
  • Whether typical manual driving performance is achieved after handover
  • At what time during the handover period is typical manual driving performance achieved
  • For how long does the driver maintain typical manual driving performance during the handover period?

The findings indicate that without a structured process there could be safety implications associated with transferring control from the autonomous system to manual driving at the speeds tested (20, 30, 40 and 50mph).

Depending on the speed of the vehicle it could have travelled a considerable distance before the driver regained typical driving performance and full control of the vehicle. Although the trial focused on planned handover, if these findings were translated into a situation where emergency handover is required, there could be further safety implications.

This needs further exploration and must consider factors such as how long the driver has been inactive, vehicle speed and road conditions.

There are also potential implications for highway network performance. VENTURER Trial 1 results revealed that the vehicle slowed during and after the handover process. This could result in a bunching effect on the network if handover is required at specific locations and vehicles (either manually driven or in autonomous mode) slow down to respond to this event. This could create a shockwave effect across the network, contributing to delays and congestion.

To mitigate these issues a structured handover process must be developed. Should industry and regulators decide that autonomous systems which require human input (such as handover) are not desirable the focus would very much be on developing fully autonomous vehicles (Level 5).

While driverless cars have the potential to realise safety benefits and improve mobility for all there are possible implications that must be considered. Among these is the potential for increased consumer demand which could have a knock-on effect on highway network capacity.

This may present a challenge in urban areas where the highway network may already be at or approaching capacity. There may also be adverse consequences for wider policy areas, such as the promotion of active travel initiatives that support general health and well-being.

VENTURER continues to inform the future direction of CAV development by creating a greater understanding of the potential opportunities and challenges on the road ahead.

Get further information on the project and read the full report

First published on Atkins Angles