Spotted: Many car companies see autonomous driving as the future of transportation. To achieve this safely, autonomous system need to be able to reason through data, rather than just react.
German company MOTOR Ai is achieving this with software that uses active inference, a model from neuroscience that allows the software to make structured, transparent driving decisions and predict the actions of surrounding traffic.
Reacting to so-called ‘edge-cases’, a situation that a model hasn’t been trained on, is a central problem in autonomous driving. To respond to these unforeseen events, most autonomous driving software relies on brute-force data collection and prediction models that use massive training datasets.
Active inference is based on the theory that living systems minimise surprise by constantly predicting sensory input and updating their beliefs about the world around them. Software based on active inference uses internal, probabilistic models to generate predictions about the environment. Because the system’s decision-making processes are rooted in its internal generative model, its outputs are more explainable and accountable compared to traditional ‘black box’ models.
MOTOR Ai’s active inference software works in this way, so it is able to adapt to new information without constant training. The company describes the software as “a quasi-dynamic, rule-based approach, which makes reproducible and verifiable decisions”.
Because of the transparency of this approach, MOTOR Ai’s autonomous driving software can be certified according to international safety standards. Its full-stack system currently meets the most stringent European and international safety and compliance requirements, including the EU AI Act and the upcoming Cyber Resilience Act provisions.
MOTOR Ai recently announced completion of a $20 million seed funding round led by Segenia Capital and eCAPITAL. This year, the company is testing vehicles using its software in road traffic throughout Germany, which are initially set to be supervised by an on board safety driver . The new capital will support MOTOR Ai on its journey to full commercial deployment, with the aim of gradually integrating the technology into regular traffic, initially for on-demand shuttle services in local public transport.