Embodied systems where perception and intelligence become action.
Robotics is one of the core topic programs in the action pillar of Neuroide. This topic covers how perception, learned models, planning, control, middleware, simulation, and validation are combined into systems that execute in the physical world.
Kinematics, dynamics, state estimation, feedback control, and numerical optimization.
Manipulation, planning, trajectory generation, collision checking, and constraints.
ROS 2, planning stacks, simulation, logging, and integration across system layers.
Benchmarking, safety, recovery behavior, and evidence under distribution shift.
Perception, Intelligence, Machine Learning, and Robotics: Toward Adaptive Embodied Systems
A cross-pillar research idea arguing that adaptive embodied systems are the right scientific unit for integrating perception, predictive intelligence, machine learning, and robotics.
A Serious Learning Path for Modern Robotics
A technical roadmap for learning robotics through dynamics, estimation, planning, perception, simulation, and software systems.
Engineering Validation for Intelligent Collaborative Robots
A systems article on closed-loop evaluation, uncertainty, benchmark design, error propagation, and adaptive safety for collaborative robots.
Robotics Tools and Software Ecosystem
A technical guide to the robotics software stack, covering ROS 2, MoveIt, RViz, OpenCV, simulation, and practical engineering workflow choices.
Robotics as a System-Level Control Problem
A systems article on intralogistics as a hierarchical control problem across estimation, allocation, planning, scheduling, execution, and recovery.
Benchmarking Open-Ended Collaborative Robot Systems
Scenario-distribution benchmark design, intervention-aware metrics, and evidence-producing protocols for collaborative robotics.
Error Propagation and Distribution Shift in Embodied Robot Systems
A systems view of compounding uncertainty, closed-loop shift, and robustness measurement across embodied robot stacks.
Evaluation-by-Design for Learning-Enabled Robotics
Metrics, evidence, scenario coverage, and intervention logic treated as part of the robot's engineering envelope.
Hybrid Learning Control for Safe Multi-Agent Intralogistics
How learning, constrained coordination, and supervisory safety fit together in multi-agent warehouse control.
Runtime-Calibrated Digital Twins for Sim-to-Real Robotics
Simulation as a runtime-calibrated experimental instrument for design, failure analysis, and sim-to-real evidence.