Eloy is just putting the last touches to our first Testbad day at UTAC Millbrook as part of the Zenzic CAM Scale-up. Our goal is to show that our artificial intelligence system can coordinate the movement of 10 vehicles across a mocked-up narrow rural road system to simulate how vehicles would pass each other on single track stretches of road.
The plan
We have selected a 1-mile loop in the city course at Millbrook, which is a lower speed part of the Testbed. Cars will travel in both directions, with 5 cars going clockwise and the other 5 cars going anti-clockwise. The 1-mile loop will include 5 coned-off sections which are only wide enough for 1 car, while the rest of the course is left as it is where 2 cars can pass each other comfortably.
UTAC Millbrook City Course
On each lap, a car is expected to interact with 10 cars coming from the other direction. Some of these interactions will be on the wider parts where they will be able to pass each other, and some will happen near or at the narrow parts. Here, one vehicle will need to stop and wait, whilst the car travelling in the opposite direction will pass through.
Simulation of the city track at UTAC
We will run repeated tests with different traffic patterns – for example allowing the cars to be spaced between 5 seconds and 1 minute apart. Some tests will be used as a baseline for efficacy analysis where we don’t have any software intervention and we can measure how the drivers navigate through the road setup.
Other multi-vehicle coordination trials
Interestingly we have seen another multi-vehicle coordination trial in the last month, performed in the United States. This test was to see if 100 vehicles could obtain benefits in a connected cruise control system.
The aim of these trials is similar to Eloy’s in that we want to show how we can reduce congestion and improve safety with connected vehicle software. The CIRCLES project aims to prevent or reduce the impact from high-density travel when sequential over braking creates phantom traffic jams.
How is Eloy different?
Eloy has taken a very different approach to introducing connected vehicle technology to the roads. We believe we need to create a more connected driver first, and allow the human driver to make slightly better decisions at the wheel. Being able to isolate smaller car interactions, such as on narrow rural roads with a few cars at a time, is far easier to manage in a human context compared to the aim of 100s of interacting cruise controls.
Eloy’s approach has 2 benefits:
Firstly, we can deploy today at scale. If the public wants the software and we are happy with the efficacy level, it can be deployed to millions of vehicles by downloading the Eloy Drive app. We can also make our software available to other applications in a short space of time.
Secondly, the narrow country lane passing places is a more realistic problem to solve because the non-participating vehicles don’t destroy the overall system effectiveness but rather just reduce it. If half the cars don’t have the software running although half the interactions will revert to a human-to-human one, the other half will still function. Importantly, cars can’t overtake each other in this setting, so if a car stops in front of another car, the second car also waits.
Hopefully Eloy’s testing this week goes well and we will report back with the early results in December!
