Self-driving tech company Wayve has launched LINGO-1, a vision-language-action model (VLAM) for self-driving to bolster learning and explainability in its AI Driver technology powering self-driving vehicles.

LINGO-1 is aimed at providing deeper insight into the decision-making and reasoning capabilities of AI models, ending the status of end-to-end AI neural nets as black boxes. Enhancing interpretability of Wayve’s AI Driver will allow the company to build safer driving intelligence for self-driving.

Trained using real-world data from Wayve’s expert drivers commentating as they drive, LINGO-1 is designed to explain the reasoning behind driving actions.

In addition to commentary, LINGO-1 can respond to questions about a diverse range of driving scenes. This feature allows Wayve to make improvements to the model through feedback.

Using LINGO-1, Wayve could incorporate natural language sources, such as the Highway Code and other safety-relevant content, to make retraining its AI models easier and simpler. By improving the raw intelligence of its AI Driver, Wayve can accelerate the learning process, enhance accuracy, and increase the technology’s capacity to handle diverse driving tasks.

Wayve is currently testing its self-driving technology on UK roads and is undertaking Europe’s largest last-mile autonomous grocery delivery trial with Asda, serving 170,000 residents across 72,000 households in London.

Alex Kendall, Co-founder & CEO of Wayve said: “LINGO-1 marks a big step for embodied AI: aligning vision, language and action to deliver more intelligent and trusted autonomous vehicles. We are excited by the capabilities we observe from LINGO-1 today and we believe natural language will provide a powerful step change in how we understand and interact with robotics.”

“At Wayve, we’re investing in pushing the boundaries of cutting-edge science to deliver a safer, smarter and more sustainable future of transport, resulting in LINGO-1. LINGO-1 opens up many possibilities for self-driving, improving the intelligence of our end-to-end AI Driver as well as bridging the gap of public trust – and this is just the beginning of maximising its potential.”