Hyperspec AI released a new tool for developers working on ADAS-enabled and autonomous vehicles (AV). The company has developed a unified platform called RoadMentor that allows users to create, train, and deploy machine learning (ML) models for real-time mapping. Hyperspec integrates the map into the ML training loop so that real-time mapping models can be developed, giving ADAS-enabled and autonomous vehicles the ability to perform outside of the HD map geofence. This expands navigable roads from less than 5% today to over 95%, for any vehicle, so the autonomous systems can learn from the ubiquitous exposure.
Today, the ML development process is fragmented with no integrated process for data collection, data management, model training, verification & validation, deployment, and fleet learning. Each step is another data transfer, leading to inefficiency and a lack of true visibility. RoadMentor enables the industry to scale through deep learning by consolidating the loop training process into one optimized infrastructure designed specifically for autonomous driving.
Autonomous driving data is largely skewed towards the highway and arterial road domains. The release of RoadMentor increases test coverage from less than 5% to over 95% across all roads, enabling edge case library build-out across the long tail of scenarios. Now ADAS functionality and autonomous driving usage and coverage can further develop allowing us to reach levels 3, 4, and 5 autonomy.
RoadMentor is offered as a freemium SaaS product, so users can process a certain amount of data at no cost. We invite developers to sign up for exclusive beta access to RoadMentor through our developer program, limited seats are available. Attendees of the International Auto Show Tech Days will be able to learn first-hand how RoadMentor improves the release cadence of autonomous driving technology development.
