Constant Time-Delay Leader Following with Neural Networks and Invariant Extended Kalman Filters for Arbitrary Trajectories
Luka Antonyshyn, Paulo Ricardo Marques de Araujo, Sidney Givigi
THE PROBLEM
This paper focuses on learning. Proposes a hybrid approach combining Seq2Seq neural networks, invariant extended Kalman filters (IEKF), and geometric Control & PlanningModel Predictive Control (MPC)A control method that repeatedly plans a short future path, acts a little, then replans. for multi-vehicle convoy following under constant time delays without inter-vehicle communication or global positioning. Validated in Simulation & Sim-to-RealSimulationA virtual environment where robots can be trained or tested. and with real robots. Read the paper by tracking the Core ConceptsTaskThe job the robot is supposed to complete, such as pick-and-place, navigation, or drawer opening. definition, the Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. or data assumptions, and the evidence that supports the claimed improvement.
HOW IT WORKS
Task framing
Core method
Data and supervision
Evaluation evidence
KEY RESULTS
This paper enables Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. vehicle convoys to follow a leader through time-delayed observations without GPS, communication, or a shared coordinate system by combining neural networks with Kalman filtering and geometric Control & PlanningModel Predictive Control (MPC)A control method that repeatedly plans a short future path, acts a little, then replans.. You get accurate Core ConceptsTrajectoryA sequence of states or actions over time. prediction and Control & PlanningControlThe method used to make the robot move the way you want. even with long delays and arbitrary leader motions.
WHY DEVELOPERS SHOULD CARE
This paper enables Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. vehicle convoys to follow a leader through time-delayed observations without GPS, communication, or a shared coordinate system by combining neural networks with Kalman filtering and geometric Control & PlanningModel Predictive Control (MPC)A control method that repeatedly plans a short future path, acts a little, then replans.. You get accurate Core ConceptsTrajectoryA sequence of states or actions over time. prediction and Control & PlanningControlThe method used to make the robot move the way you want. even with long delays and arbitrary leader motions.
LIMITATIONS
The main limitation to check is whether the claimed behavior holds outside the paper's reported setup. That means testing across different Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments, scenes, objects, and data distributions.
WHAT COMES NEXT
The practical next step is independent reproduction with clear baselines, ablations, and stress tests. For a developer, the useful follow-up is to map the paper's learning assumptions onto a concrete Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. stack, then test the smallest version of the method that could run end to end.