Sign up for our weekly emails to stay up to date on the latest news!
Wayve, a leader in Embodied AI for self-driving technologies, has introduced PRISM-1, a novel 4D reconstruction model that significantly enhances the testing and training of its ADAS and autonomous driving technology. PRISM-1 represents a major advancement in 4D reconstruction, enabling scalable, realistic resimulations of complex driving scenes with minimal engineering or labeling input.
Key Highlights:
- PRISM-1 utilizes camera-only inputs, avoiding traditional reliance on LiDAR and 3D bounding boxes.
- Reconstructs dynamic elements such as cyclists, pedestrians, brake lights, and road debris.
- Launch accompanied by WayveScenes101 dataset, supporting research in novel view synthesis models for driving.
First showcased in December 2023 through the Ghost Gym neural simulator, PRISM-1 employs novel view synthesis to create precise 4D scene reconstructions (3D in space plus time) using only camera inputs. This method promises to revolutionize simulation for autonomous driving by accurately and efficiently simulating the dynamics of complex and unstructured real-world environments. Unlike traditional methods that rely on LiDAR and 3D bounding boxes, PRISM-1 utilizes novel view synthesis techniques to accurately depict moving elements such as pedestrians, cyclists, vehicles, and traffic lights, including detailed features like clothing patterns, brake lights, and windshield wipers.
Achieving realism is critical for building an effective training simulator and evaluating driving technologies. Traditional simulation technologies treat vehicles as rigid entities and fail to capture safety-critical dynamic behaviors like indicator lights or sudden braking. PRISM-1, however, uses a flexible framework that excels at identifying and tracking changes in the appearance of scene elements over time, enabling it to precisely resimulate complex dynamic scenarios with elements that change in shape and move throughout the scene. It distinguishes between static and dynamic elements in a self-supervised manner, avoiding the need for explicit labels, scene graphs, and bounding boxes to define the configuration of a busy street. This approach maintains efficiency, even as scene complexity increases, ensuring that more complex scenarios do not require additional engineering effort. This makes PRISM-1 a scalable and efficient solution for simulating complex urban environments.
Jamie Shotton, Chief Scientist at Wayve, remarked, “PRISM-1 bridges the gap between the real world and our simulator. By enhancing our simulation platform with accurate dynamic representations, Wayve can extensively test, validate, and fine-tune our AI models at scale.
“We are building Embodied AI technology that generalizes and scales. To achieve this, we continue to advance our end-to-end AI capabilities, not only in our driving models but also through enabling technologies like PRISM-1. We are also excited to publicly release our WayveScenes101 dataset, developed in conjunction with PRISM-1, to foster more innovation and research in novel view synthesis for driving.”
Alongside the launch of PRISM-1, Wayve is also releasing its WayveScenes101 Benchmark, a dataset comprising 101 diverse driving scenarios from the UK and US, including urban, suburban, and highway scenes under various weather and lighting conditions. Wayve aims for this dataset to support the AI research community in advancing novel view synthesis models and developing more robust and accurate scene representation models for driving.
Sign up for our weekly emails to stay up to date on the latest news!
