Anca D. Dragan

About

Anca D. Dragan is a pioneering roboticist whose research sits at the intersection of human-robot interaction, motion planning, and value alignment. Based at UC Berkeley, she has made foundational contributions to how robots understand, predict, and communicate with humans in collaborative settings. Dragan's early work on CHOMP (738 citations) established a landmark approach to trajectory optimization, using covariant gradient techniques to produce smooth, efficient robot motion. Her influential research on legibility and predictability (515+ citations) formalized a crucial distinction: robots should generate motion that actively communicates intent, not merely motion that is technically correct. This insight reshaped how researchers think about expressive robot behavior. Her work on shared control (373 citations) and seamless handovers (257 citations) extended these principles to real-world collaboration, while her contributions to autonomous driving (425 citations) challenged the field to move beyond passive prediction toward robots that strategically account for their influence on human behavior. Collaborating on Cooperative Inverse Reinforcement Learning (326 citations), she helped lay theoretical groundwork for value alignment — ensuring AI systems genuinely serve human interests. Her preference-based learning framework (259 citations) further democratized reward specification for complex robotic systems. Collectively, Dragan's work has accumulated thousands of citations, cementing her as one of the most influential voices shaping socially intelligent, human-centered robotics.

Research Focus

Key Achievements

Top Papers

1. 1
   CHOMP: Covariant Hamiltonian optimization for motion planning

   738 citations · 2013
2. 2
   Legibility and predictability of robot motion

   515 citations · 2013
3. 3
   Planning for Autonomous Cars that Leverage Effects on Human Actions

   425 citations · 2016
4. 4
   A policy-blending formalism for shared control

   373 citations · 2013
5. 5
   Legibility and predictability of robot motion

   349 citations · 2013
6. 6
   Cooperative Inverse Reinforcement Learning

   326 citations · 2016
7. 7
   Active Preference-Based Learning of Reward Functions

   259 citations · 2017
8. 8
   Towards Seamless Human-Robot Handovers

   257 citations · 2013
9. 9
   Effects of Robot Motion on Human-Robot Collaboration

   245 citations · 2015
10. 10
    Planning for cars that coordinate with people: leveraging effects on human actions for planning and active information gathering over human internal state

    180 citations · 2018

Key Collaborators

Contact & Links