Job DescriptionKey Responsibilities:Reinforcement Learning for ManipulationDesign and implement deep reinforcement learning (RL) policies for robotic manipulation in dynamic settingsDevelop self-learning and policy optimization techniques to improve decision-makingTrain inverse kinematics (IK) models for real-time, adaptive motion controlDeep Learning for Motor Control & DexterityBuild neural network-based control models for grip compliance, force adaptation, and fluid motionUse transformer models for intelligent motion sequencingDevelop Sim2Real pipelines to transfer trained models to physical robotsMotion Planning & Collision AvoidanceImplement and refine trajectory planning using RRT, PRM, Hybrid-A*, TEB*Integrate motion control policies with ROS2 MoveIt! and OrocosEnable grasping strategies that adapt to force and handle unstructured environmentsSensor Fusion & Environment MappingBuild systems combining data from LiDAR, depth cameras, IMU, and force sensorsUse Neural SLAM techniques for accurate mapping and object manipulationExplore Vision-Language Models (VLMs) to support semantic understanding in robotic tasksTesting, Simulation & DeploymentBenchmark model performance against real-world scenariosTroubleshoot and refine control pipelines for reliabilityDevelop frameworks for Sim2Real validation and deploymentDocumentation & ResearchMaintain clear and detailed documentation of models, training processes, and system designStay updated on research in AI, robotics manipulation, and autonomous control systemsMust-Have Skills:Strong foundation in Reinforcement Learning, Deep Learning, and trajectory optimizationExperience with ROS2 MoveIt!, Orocos, NVIDIA Isaac Sim, Groot, and OmniverseHands-on work in Sim2Real transfer and AI-based robotic controlFamiliarity with motion planning algorithms, sensor fusion, and SLAM frameworks
