Justin Kerr
Justin Kerr researches robot perception, three-dimensional scene understanding and manipulation. His work includes a 2022 system for untangling long cables and the 2023 LERF method, which connects language embeddings with neural scene models. He also coauthored Nerfstudio and work on grouping objects in three dimensions. As of October 2026, Kerr is a Berkeley PhD student advised by Angjoo Kanazawa and Ken Goldberg, and a Nerfstudio maintainer.
Timeline
-
2022
Untangling cables
Kerr coauthored a system for untangling long cables using a two-arm robot. The team developed motion actions and grippers for managing cable slack. Its algorithm combined these actions to handle knots and crossings in tested configurations.
-
2023
Connecting language with 3D space
With collaborators, Kerr introduced Language Embedded Radiance Fields, or LERF. The method places language embeddings from CLIP inside a neural radiance field. It supports open-ended language queries about locations in a reconstructed scene.
-
2023–2024
Tools and grouping in 3D
His work includes Nerfstudio, a modular framework for neural radiance fields, in 2023. His work also includes GARField in 2024, which learns hierarchical groups in three-dimensional scenes using a scale-conditioned affinity field.
-
As of October 2026
Robot perception research
Kerr is a Berkeley AI Research PhD student advised by Angjoo Kanazawa and Ken Goldberg. He describes work on perception for manipulation and active vision. He also maintains Nerfstudio.
Sources and credits
Photo: kerrj.github.io; Justin Kerr profile photograph; creator not stated. License: Not stated by source. Source: kerrj.github.io. Cropped to a square and resized. Ask us to remove or replace this photo.