Sanjit A. Seshia

About

Sanjit A. Seshia is a prominent computer scientist whose research spans cyber-physical systems, formal methods, autonomous systems, and human-robot interaction. His landmark textbook, *Introduction to Embedded Systems: A Cyber-Physical Systems Approach* (2013), has become a foundational resource in the field, amassing over 1,179 citations and shaping how engineers understand the intersection of computation and physical processes. Seshia has made significant contributions to autonomous vehicle planning, particularly in developing frameworks where self-driving cars actively anticipate and influence the behavior of human drivers, rather than merely reacting to them — work that has collectively garnered hundreds of citations. His research on preference-based reward learning addresses the critical challenge of efficiently encoding human intentions into robotic and autonomous systems. Beyond autonomy, Seshia has advanced formal verification methods for multi-robot systems, introducing SMT-based compositional planning and assumption mining for reactive system synthesis. His development of Scenic, a probabilistic programming language for designing and stress-testing cyber-physical and machine-learning-based systems, reflects his commitment to rigorous, safety-conscious engineering. Across these contributions, Seshia's work consistently bridges theoretical rigor with practical impact in safety-critical autonomous systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Introduction to Embedded Systems - A Cyber-Physical Systems Approach

   1,179 citations · 2013
2. 2
   Planning for Autonomous Cars that Leverage Effects on Human Actions

   425 citations · 2016
3. 3
   Active Preference-Based Learning of Reward Functions

   259 citations · 2017
4. 4
   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
5. 5
   Information gathering actions over human internal state

   165 citations · 2016
6. 6
   Automated composition of motion primitives for multi-robot systems from safe LTL specifications

   127 citations · 2014
7. 7
   Mining assumptions for synthesis

   102 citations · 2011
8. 8
   Scenic: a language for scenario specification and data generation

   83 citations · 2022
9. 9
   Combining Model Checking and Runtime Verification for Safe Robotics

   75 citations · 2017
10. 10
    DRONA

    73 citations · 2017

Key Collaborators

Contact & Links