New research from the University of Munich may have developed an artificial intelligence system which can warn drivers of potentially serious incidents seven seconds in advance.
The system which has been analysed alongside BMW could prove a vital tool as more and more vehicle manufacturers look at autonomous driving. The biggest drawback for many AI in vehicles currently is that they can only adjust to a limited range of scenarios, meaning that more often than not a driver will have to take controls and take action for safety reasons.
But the Munich School of Robotics and Machine Intelligence (MSRM) may have found a solution, with technology that learns from previous situations, particularly when human intervention is required.
Like many assisted driving technologies, the MSRM uses sensors and cameras to capture data from the vehicles immediate environment, aspects such as road conditions, weather, visibility and based on a recurrent neural network can learn to recognise patterns. Where a pattern occurs that is not recognised, the warning system can call for human intervention.
“To make vehicles more autonomous, many existing methods study what the cars now understand about traffic and then try to improve the models used by them. The big advantage of our technology: we completely ignore what the car thinks. Instead, we limit ourselves to the data based on what actually happens and look for patterns," said Professor Eckehard Steinbach, Chair of Media Technology at the university.
“In this way, the AI discovers potentially critical situations that models may not be capable of recognising, or have yet to discover. Our system therefore offers a safety function that knows when and where the cars have weaknesses."
The MSRM has already been put to the test with BMW’s autonomous vehicles on public roads, with 2500 situations which required human intervention. This data has been used to predict critical situations with 85 per cent accuracy - up to seven seconds before they occur.