Peter Carr

About

Peter Carr is a leading researcher in autonomous vehicle perception and intelligent camera systems. His most impactful contribution is the creation of **Argoverse**, a landmark dataset for 3D tracking and motion forecasting that has garnered over 1,400 citations. This work provides richly annotated sensor data from autonomous fleets, including detailed 3D tracking and map information, which has become a critical benchmark for advancing self-driving technology. Beyond autonomous vehicles, Carr has pioneered novel approaches for **autonomous cinematography**, developing hybrid robotic/virtual camera controllers that mimic human operators to produce aesthetic, professional-quality video for sports and event recording. His data-driven methods for robotic pan-tilt-zoom cameras, which learn optimal framing without predefined heuristics, represent a significant advance in automated broadcasting. Earlier in his career, Carr explored distributed imaging systems inspired by biological compound eyes, investigating how networked cameras can capture wide field-of-view scenes. Through his work on Argoverse and intelligent camera systems, Carr has made foundational contributions that bridge computer vision, robotics, and autonomous systems, providing essential tools and methodologies for researchers and engineers working on real-world perception challenges.

Research Focus

Key Achievements

Top Papers

1. 1
   Argoverse: 3D Tracking and Forecasting With Rich Maps

   1,420 citations · 2019
2. 2
   Argoverse: 3D Tracking and Forecasting with Rich Maps

   157 citations · 2019
3. 3
   Hybrid robotic/virtual pan-tilt-zom cameras for autonomous event recording

   68 citations · 2013
4. 4
   Mimicking Human Camera Operators

   34 citations · 2015
5. 5
   <title>Performance benefits and limitations of a camera network</title>

   4 citations · 2005
6. 6
   Distributed imaging using compound eye sensors

   2 citations · 2004

Key Collaborators

Contact & Links