
Blending MPC Value Function Approximation for Efficient Reinforcement Learning
ModelPredictive Control (MPC) is a powerful tool for controlling comple...
read it

Declarative vs Rulebased Control for Flocking Dynamics
The popularity of rulebased flocking models, such as Reynolds' classic ...
read it

The Value of Planning for InfiniteHorizon Model Predictive Control
Model Predictive Control (MPC) is a classic tool for optimal control of ...
read it

Neural Lyapunov Model Predictive Control
This paper presents Neural Lyapunov MPC, an algorithm to alternately tra...
read it

Optimal Cost Design for Model Predictive Control
Many robotics domains use some form of nonconvex model predictive contro...
read it

Information Theoretic Model Predictive QLearning
Modelfree Reinforcement Learning (RL) algorithms work well in sequentia...
read it

L_1regularized Boltzmann machine learning using majorizer minimization
We propose an inference method to estimate sparse interactions and biase...
read it
Practical Reinforcement Learning For MPC: Learning from sparse objectives in under an hour on a real robot
Model Predictive Control (MPC) is a powerful control technique that handles constraints, takes the system's dynamics into account, and optimizes for a given cost function. In practice, however, it often requires an expert to craft and tune this cost function and find tradeoffs between different state penalties to satisfy simple high level objectives. In this paper, we use Reinforcement Learning and in particular value learning to approximate the value function given only high level objectives, which can be sparse and binary. Building upon previous works, we present improvements that allowed us to successfully deploy the method on a real world unmanned ground vehicle. Our experiments show that our method can learn the cost function from scratch and without human intervention, while reaching a performance level similar to that of an experttuned MPC. We perform a quantitative comparison of these methods with standard MPC approaches both in simulation and on the real robot.
READ FULL TEXT
Comments
There are no comments yet.