Autonomous Systems that Model Human Behaviour

Autonomous Vehicles need to be tested in simulation before trying them on real roads. For these simulations to be lifelike, they must include humans displaying realistic behaviours in addition to simulating the road environment accurately. We build realistic A.I.-based human behaviour models, using a machine learning technique called Imitation Learning. Our models enable scalable, fast, robust, and safe testing of Autonomous Vehicles.

Request a demo

How We Do It

There are two stages to our learning process: extraction of behaviours from raw data, and learning to imitate those behaviours. Computer Vision detects road users and tracks their motion. Imitation Learning, also known as Learning from Demonstration, then learns to create new, artificial behaviours which are indistinguishable from the ones used as the demonstration input, meaning our virtual humans are completely realistic. Our virtual humans integrate with our customers’ preferred simulator via a standard API.

Close video


We learn the bad human behaviours as well as the good, as it’s essential to test an AV in simulation against the kinds of scenes it will encounter in the real world. More realistic virtual testing reduces the need for costly and risky physical testing at an early stage.


Our virtual humans support highly scalable testing of AVs. Efficient compute runs hundreds of virtual humans on the equivalent of a standard laptop, and rare edge cases can be tested repeatedly. Scalable, automated testing speeds up software development.

Easy Integration

Our virtual humans are platform-agnostic. They integrate into all the common simulation engines, either as a Library or through a co-simulation, using standard frameworks for agent-control modules.

How safe is a self-driving car?

This is the perhaps the biggest question facing AV developers, insurers and regulators.
We have been awarded two multi-million-£ projects by Innovate UK, the government’s innovation agency, to build a simulation system that can help answer just this question.

In the OmniCAV project, which we are leading, we are building a market-first simulation of rural, highway and urban roads that can be used to certify an AV as safe.

Partners: Latent Logic (Lead), Admiral, Arcadis, Aimsun, Arrival, Ordnance Survey, Oxfordshire County Council, RACE, XPI Simulation, WMG

The VeriCAV project is building a smart simulation which dynamically changes its test conditions based on the AV’s performance, so the AV gets tested more on the scenarios it finds more challenging.

Partners: Horiba Mira (Lead), Aimsun, Latent Logic, Propelmee, SAIC, Transport Systems Catapult, University of Leeds

Explore applications


AI. Humanised

Our Mission

The idea for our company originated from our research work developing interactive tele-presence robots for elderly care homes, improving the quality of care and sense of community. We believe huge societal and commercial value will be created by autonomous systems that co-exist in harmony with humans. Our technology moves us closer to that goal.

Learn About Us

Work With Us

We make state-of-the-art AI that works in the real world. Working with us means exposure to the latest deep learning research and the opportunity to transfer that thinking out of the lab into software that brings our customers tangible benefits. We are building a place where people are free to think and create in the way that works best for them.

Join us