Safety has always been the ultimate concern in the field of autonomous vehicles, and aiming to address these concerns, software provider Oxbotica is using advanced artificial intelligence (AI) in the metaverse to accelerate the safe and efficient deployment of autonomous vehicle technology, while also reducing carbon emissions generated by vehicles when driving.
The launch comes just weeks after Oxbotica announced that it had completed the first safe and sustainable deployment of a zero-occupancy, fully autonomous, new-type electric vehicle on publicly accessible roads in Europe.
The new Oxbotica MetaDriver suite of tools includes virtual world simulation, automated discovery of challenging scenarios, and real-time data expansion. It is designed to help autonomous vehicles answer three core questions: “Do I see it right?”, “Am I doing it right?”, and “Can I trust myself right now?”
MetaDriver generates what the company says is a “vast” bank of scenarios that are used to test and refine autonomous vehicle operations and behaviours without ever needing to physically drive in them, accelerating commercial readiness. The technology also provides the ability to build a digital twin representation from real-world-gathered data.
At the heart of the system is AI which has learned to automatically seek out what Oxbotica calls rare, unusual and unseen scenarios, known as “edge cases”. This allows both the testing of autonomy systems against “corner cases” and for the systems to learn to handle them. MetaDriver is claimed to offer edge-case discovery rates at an average of 1,000 times faster than traditional testing methods – up to 35,000 times faster in some situations.
Oxbotica said that by using MetaDriver’s data expansion, each “meta mile” can be tested in all conditions, including rain, snow, fog, at night or in the day, eliminating the need to wait for various weather conditions to appear during real-world testing. The company claimed that the capabilities of MetaDriver enable fast, targeted and diverse testing, which it said will give autonomous vehicle operators confidence that their vehicles are equipped to operate safely, accurately and effectively in their specialised environments, whether that is a city, mine, wind farm, solar park or airport.
“Oxbotica MetaDriver is a suite of tools that offers so much more than straightforward simulation,” said Oxbotica founder and CTO Paul Newman. “Using the metaverse in this way provides us and our customers with practically unlimited test challenges, revolutionising how we are accelerating autonomous vehicle commercial deployment to make the Earth move.
“MetaDriver allows us to do the hard miles without the conventional need to physically drive the miles. The autonomous vehicle industry has become over-indexed on the number of miles travelled as a synonym for safety, but what we actually care about is coverage of hard cases, not the endless miles endured. Oxbotica MetaDriver addresses this in a new and exciting way.”
Read more about autonomous vehicles
- Swedish researcher cuts through the hype around autonomous vehicles, with Linköping University’s Michael Felsberg making realistic predictions about when self-driving cars will become available.
- Verizon and Cisco partner on proof-of-concept to advance autonomous vehicle tech with streamlined architecture based on mobile edge compute, LTE connectivity and industrial IoT to rearchitect autonomous driving by virtualising roadside infrastructure to support low-latency communications.
- Ruggedised storage takes autonomous vehicle research to edge as Mississippi State University’s Center for Advanced Vehicular Systems selects Quantum Corp’s R-Series SSDs to collect sensor data at the edge.
Original post: https://www.computerweekly.com/news/252521611/Oxbotica-taps-metaverse-to-improve-autonomous-vehicle-detection-scenarios