The development of autonomous vehicles (AVs) is still in its infancy. How do OEMs test these vehicles to ensure they are safe on the road in the real world?  AB Dynamics has been developing testing systems for automakers for decades, pushing back the boundaries of advanced driver assistance (ADAS) test systems, the building blocks of an AV. To learn more about how the auto industry is developing testing regimes for AVs, we spoke to Jeremy Ash, Head of Commercial, AB Dynamics.

We are hearing a lot about autonomous vehicles. In what ways can AB Dynamics help test these vehicles?

The move towards autonomous vehicles is certainly one of the most significant trends in the automotive industry to-date. As a company, we have designed a breadth of test equipment to help those developing this technology. From testing and fine-tuning suspension parameters through to studying human behaviour during a ‘handover’ procedure, our expertise is diverse. Fundamentally, our products fall into two categories: track and laboratory. The track equipment encompasses our driving robots, ADAS targets and data capture equipment while the lab side covers the advanced Vehicle Driving Simulator (aVDS), Kinematics and Compliance machine (SPMM 5000) and Steering System Test Machine (SSTM).

To give a general example, our equipment can be used to acquire the parameters for a digital model. This model can then be used in virtual simulations or in a driving simulator like the aVDS, with a driver in the loop. The results obtained during the simulation can then be verified by closely replicating the tested scenarios at a test track or proving ground, using a real vehicle.

Our ability to choreograph real vehicles and soft targets allows us to create complex traffic scenarios on the test track.

Our ability to choreograph real vehicles and soft targets allows us to create complex traffic scenarios on the test track. Our customers can use this to recreate and delve deeper into situations they may have observed while testing on the public roads.

We understand that AB Dynamics has developed a suite of ADAS testing equipment, including LaunchPad. Could you tell us a little more?

Our ADAS test equipment is used across the globe by vehicle manufacturers, test houses and Tier 1s looking to generate accurate and actionable data from controlled testing. The suite of equipment we produce is intended to make this testing as simple as possible, with scenarios from simulation easily replicated in the real world at a test track or proving ground.

LaunchPad is a motorised platform used to safely propel dummies which represent pedestrians, cyclists and other vulnerable road users (VRUs) in ADAS testing. We can replicate the speed and movement of VRUs to examine how an autonomous car or driver assistance system reacts to, say, a child stepping out in front of the vehicle. Not only do these targets represent VRUs visually, they also provide an accurate radar profile. LaunchPad is complemented by the Guided Soft Target (GST), which follows the same basic principle but carries a full dummy vehicle made of foam. The GST has already seen extensive use with safety organisations like Euro NCAP and NHTSA, to test the effectiveness of advanced driver-assistance systems like autonomous emergency braking (AEB).

The vehicle under test (VUT) may be controlled by our systems (either driving robots or by wire) or may be human-driven or fully autonomous. In each case, accurate synchronisation to the VUT is possible thanks to our unique Synchro technology.

We understand that AB Dynamics is demonstrating self-driving motorbikes to simulate real-world driving conditions for autonomous vehicles (AVs). How is that side of your business shaping up and what stage have you reached?

The self-driving motorcycle that we demonstrated last year is now in the process of being commercialised. The project came about after several customers expressed concerns about their systems being able to identify fast-moving motorcycles. It is one thing to detect cars, buses and trucks at regular speeds, it is entirely different to try and track the rapid agility and sometimes unpredictable behaviour of motorcycles. They can ‘appear’ quite rapidly and in many places are now legally allowed to filter past stationary or slow-moving traffic, complicating things further. Motorcyclists are tragically overrepresented in road fatality statistics, so our aim is to ensure that engineers have the tools they need to accurately represent them during sensor, system and vehicle development. We are currently refining the system after a number of successful trials, with the hope that it can soon be used to create more representative street scenarios for AV testing.

When testing AVs, I guess certain situations will always be difficult to predict and prepare for. Mopeds weaving around slow-moving traffic is one thing (above) yet the behaviour of animals on a highway is another. How do you address that?

An AV can encounter an infinite number of potential scenarios, making testing complex.

An AV can encounter an infinite number of potential scenarios, making testing complex. The approach we are seeing our customers take is to try and find the best combination of testing across public roads and proving grounds, and to combine this with extensive simulation work. Our LaunchPad could be used to move dummy animals in a realistic fashion – it isn’t limited to straight line trajectories – so that an AV’s safety systems could be assessed in a variety of wildlife encounters.

What is your vision of how the industry is likely to develop testing regimes for AVs? i.e. do you see future legislation and vehicle safety testing requiring ADAS systems and AVs to be validated in increasingly complex scenarios?

This is a complex question! We see a wide variation in how testing is being carried out at present, and the approach of traditional OEMs is often quite different from how a younger tech start-up would set out to do testing. We’ve seen some bold predictions about when you will be able to buy a driverless car, but before this can happen there is a huge amount of work to do in designing and implementing the means to prove that AVs are safe. We work with various testing and regulatory bodies around the world and most of them are exploring this question but we certainly haven’t seen any final protocols for how this will be done.

If safety systems become sufficiently good in an autonomous vehicle, could crumple zones and seat belts be eliminated from cars?

That’s an interesting question. You don’t wear a seat belt on a train, but they operate in a much more restricted and predictable environment. Even with a perfect self-driving car, if someone runs out in front of you then it will cause a fairly violent brake input, and if you weren’t wearing a seat-belt then you’d likely be hurt. There are many unpredictable external factors that mean removing passenger safety measures is unlikely, so I don’t expect this to happen any time soon.