Embedded Files
Current Research
Current Research
Supervise Learning from Demonstrations using LLMs
Supervise Learning from Demonstrations using LLMs
Fine-tuning to unseen data using online policy rollouts and LLMs to supervise learning online.
Prompt: Evaluate robot trajectory (Red)
Learning Implicit Social Navigation Behavior using Deep Inverse Reinforcement Learning
Learning Implicit Social Navigation Behavior using Deep Inverse Reinforcement Learning
Learning to avoid deadlocks and navigate around humans in the scene.
Multi-modal Human-Robot Interaction design for increased engagement using Socially Assistive Robots (SARs)
Multi-modal Human-Robot Interaction design for increased engagement using Socially Assistive Robots (SARs)
Our robot assists the people for a special FUN exhibit at the University of Michigan Museum of Art (UMMA).
We designed an app that runs on an iPad to display different artworks and use a GPT-based agent that can answer questions about these artworks on display.
Particle RRT with Informed Replanning under uncertainty
Particle RRT with Informed Replanning under uncertainty
Proposed a particle RRT where the different particles were used to characterize the uncertainty in the robot model and do re-planning when the maximum likelihood particle trajectory was far from planned trajectory.
Motion Planning using a PR2 Robot in OpenRAVE
Motion Planning using a PR2 Robot in OpenRAVE
Completed discrete (A* and ANA*) and Sampling based planning (RRT and Bi RRT) for planning the motion of PR2 robot in OpenRAVE in a cluttered environment
Approximate POMDP in a grid world
Approximate POMDP in a grid world
Completed implementation of Augmented MDP, MC-POMDP (approximate POMDP) for a grid world in simulation toy problem.
Honda R&D Americas Internship Project
Honda R&D Americas Internship Project
Develop a Control Algorithm to carry out the longitudinal control of the vehicle for a custom application. Plan the trajectory of the vehicle for L2 autonomy, using a pure pursuit algorithm, using GPS data for feedback.
Robotic Systems Lab Course
Robotic Systems Lab Course
Designed a PID-controller for the balancing motion of a two wheeled robot and drive it autonomously in a maze captured by a motion capture system on a path planned using Virtual Potential Field
Control of Three Link Manipulator using Fractional Order PID Controller
Control of Three Link Manipulator using Fractional Order PID Controller
Reports research on the control of three link planar manipulator system to evaluate the performance of Fractional Order PID against Integer Order PID for robustness in control.
Study of Optimization Methods for Tuning of PID Gains for Three Link Manipulator
Study of Optimization Methods for Tuning of PID Gains for Three Link Manipulator
Reported different heuristic optimization techniques for trajectory tracking and disturbance rejection in a three link manipulator system.
Page updated
Report abuse