Prof. Jason Corso

He is the Toyota Professor of AI & Robotics at the University of Michigan, where he is also a Professor of Robotics and Electrical Engineering and Computer Science. His research examines how intelligent systems perceive, understand, and reason about complex real-world environments, with work spanning robot perception, video understanding, vision-language systems, activity recognition, multimodal learning, and artificial intelligence for healthcare. (Michigan Robotics)

Prof. Corso is also the Co-Founder and Chief Science Officer of Voxel51, the Physical AI data company behind FiftyOne. Founded from research at the University of Michigan, Voxel51 has become a major platform for teams developing computer vision, autonomous vehicles, robotics, and other systems that must interpret the physical world. (Voxel51)

FiftyOne gives AI teams a unified environment for exploring multimodal data, evaluating models, finding failure cases, improving annotations, curating high-value training data, and understanding how systems perform outside controlled laboratory conditions. It supports images, video, LiDAR, radar, IMU, audio, geolocation, text, three-dimensional data, and synchronized time-series sensor information. (Voxel51)

Voxel51 technology is used by organizations including General Motors, Sony Honda Mobility, Porsche, Bosch, Google, and other leading teams working across autonomous vehicles, robotics, automotive systems, manufacturing, healthcare, security, and defense. These organizations use FiftyOne to manage large and complex datasets, investigate model behavior, locate rare scenarios, and accelerate the development of production-grade Physical AI. (Voxel51)

For autonomous-vehicle teams, this work addresses one of the industry's most difficult problems: finding the relatively rare situations that can determine whether a system is genuinely safe. These include occluded pedestrians, unprotected left turns, construction zones, sensor glare, unusual road behavior, and other long-tail events hidden within enormous volumes of fleet data. FiftyOne enables teams to search for these scenarios across petabyte-scale datasets and use them to improve perception, planning, and end-to-end driving models. (Voxel51)

Prof. Corso's work is distinguished by the connection between foundational research and deployed AI. His research has advanced video segmentation, activity recognition, video-to-text systems, structured prediction, multimodal reasoning, robotic perception, and methods for extracting useful knowledge from large collections of visual data. Through Voxel51, that research has been translated into infrastructure used by organizations building real-world AI systems.

His relationship with Outonomous also has deep roots. Omar Mukhtar was a student in Prof. Corso's computer vision course early in Corso's faculty career. After examining Outonomous' work, Prof. Corso publicly highlighted both the scale problem facing autonomous-vehicle adoption and the potential of Outonomous' generalizable, reversible, plug-and-play approach.

In "Will It Really Take 4,000 Years?", Prof. Corso examined the enormous gap between the approximately 1.6 billion vehicles operating globally and the comparatively tiny number of autonomous vehicles deployed. He argued that the human, safety, and financial costs of road accidents make it critical to accelerate AV adoption, and identified Outonomous' lower-cost, scalable approach as a promising response to that challenge. Read the post on LinkedIn

In his follow-up, "Plug-and-Play AVs?", Prof. Corso described Outonomous as a generalizable and reversible autonomy platform combining LiDAR and camera fusion with a modular installation approach. He emphasized its ability to upgrade vehicles across model years through a transferable installation completed in approximately 30 minutes. He also recognized Omar as an early pioneer in bringing AI into practice, beginning with his first complete AI product in 2004 while living in a developing country. Read the post on LinkedIn

At Outonomous, Prof. Corso brings extraordinary depth in computer vision, robotics, multimodal AI, autonomous systems, data infrastructure, and the evaluation of AI operating in complex physical environments. His work directly aligns with Outonomous' mission to make safer autonomous technology deployable across the enormous installed base of vehicles already on the road.

• University of Michigan Profile
• Voxel51
• FiftyOne
• Voxel51 for Autonomous Vehicles
• Google Scholar
• Research Website
• Will It Really Take 4,000 Years?
• Plug-and-Play AVs?