Byron Boots

About

Byron Boots is a prominent researcher at the intersection of machine learning, robotics, and sequential decision-making. His work spans probabilistic modeling, motion planning, imitation learning, and robot perception, with a particular focus on developing principled, efficient algorithms that bridge theoretical rigor and real-world applicability. Boots has made significant contributions to the spectral learning of sequential models, including foundational work on Reduced-Rank Hidden Markov Models and Hilbert Space Embeddings of HMMs, which together have garnered over 280 citations and offer powerful alternatives to traditional expectation-maximization approaches. His research on predictive state representations (164 citations) tackled the critical challenge of closing the loop between learning and planning in partially observable environments. In robotics, Boots pioneered the use of Gaussian processes for motion planning, producing two highly influential works (combined 256 citations) that reframe trajectory optimization as probabilistic inference, enabling faster and more flexible planning for high degree-of-freedom robots. His work on imitation learning through Deeply AggreVaTeD (92 citations) demonstrated how differentiable deep learning models can leverage near-optimal oracles for sequential prediction tasks. Collectively, Boots' research has shaped modern approaches to robot learning and planning, making him a key figure in advancing intelligent, adaptive robotic systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Hilbert Space Embeddings of Hidden Markov Models

   177 citations · 2018
2. 2
   Closing the learning-planning loop with predictive state representations

   164 citations · 2011
3. 3
   Gaussian Process Motion planning

   141 citations · 2016
4. 4
   Motion Planning as Probabilistic Inference using Gaussian Processes and Factor Graphs

   115 citations · 2016
5. 5
   Reduced-Rank Hidden Markov Models

   106 citations · 2009
6. 6
   Deeply AggreVaTeD: Differentiable Imitation Learning for Sequential\n Prediction

   92 citations · 2017
7. 7
   Simulation-based design of dynamic controllers for humanoid balancing

   45 citations · 2016
8. 8
   Space-time functional gradient optimization for motion planning

   37 citations · 2014
9. 9
   Incremental sparse GP regression for continuous-time trajectory estimation and mapping

   36 citations · 2016
10. 10
    Learning predictive models of a depth camera & manipulator from raw execution traces

    35 citations · 2014

Key Collaborators

Contact & Links