• Publications
  • Influence
Parallelizing Exploration-Exploitation Tradeoffs with Gaussian Process Bandit Optimization
This work develops GP-BUCB, a principled algorithm for choosing batches, based on the GP-UCB algorithm for sequential GP optimization, and proves a surprising result; as compared to the sequential approach, the cumulative regret of the parallel algorithm only increases by a constant factor independent of the batch size B.
Spike detection using the continuous wavelet transform
This paper combines wavelet transforms with basic detection theory to develop a new unsupervised method for robustly detecting and localizing spikes in noisy neural recordings that exceeds other commonly used methods in a wide variety of recording conditions.
The explicit dynamic model and inertial parameters of the PUMA 560 arm
To provide COSMOS, a dynamic model based manipulator control system, with an improved dynamic model, a PUMA 560 arm was disassembled; the inertial properties of the individual links were measured;
A modal approach to hyper-redundant manipulator kinematics
This paper presents novel and efficient kinematic modeling techniques for "hyper-redundant" robots based on a "backbone curve" that captures the robot's macroscopic geometric features and introduces a "modal" approach, in which a set of intrinsic backbone curve shape functions are restricted to a modal form.
Sensor-Based Exploration: The Hierarchical Generalized Voronoi Graph
This paper defines the HGVG structure: a robot can plan a path between two locations in its work space or configuration space by simply planning a path onto the HNVG, then along the HCVG, and finally from theHGVG to the goal.
Weighted line fitting algorithms for mobile robot map building and efficient data representation
An algorithm to find the line-based map that best fits sets of two-dimensional range scan data via maximum likelihood formalism, which weights each point's influence on the fit according to its uncertainty, which is derived from sensor noise models.
Weighted range sensor matching algorithms for mobile robot displacement estimation
A "weighted" matching algorithm to estimate a robot's planar displacement by matching two-dimensional range scans and develops uncertainty models that account for effects such as measurement noise, sensor incidence angle, and correspondence error.
End-to-End Safe Reinforcement Learning through Barrier Functions for Safety-Critical Continuous Control Tasks
This work proposes a controller architecture that combines a model-free RL-based controller with model-based controllers utilizing control barrier functions (CBFs) and on-line learning of the unknown system dynamics, in order to ensure safety during learning.
Implications of Assist-As-Needed Robotic Step Training after a Complete Spinal Cord Injury on Intrinsic Strategies of Motor Learning
Results indicate that the window training approach, which includes loose alternating interlimb coordination, is more effective than a fixed trajectory paradigm with rigid alternating inter Limb coordination or an AAN paradigm without any interlimB constraints in promoting robust postinjury stepping behavior.