Reinforcement Learning | LAMARR

Guided Reinforcement Learning – Application to three Robotic Tasks

July 26th, 2023

Julian

Eßer

Guided RL aims at learning control policies for real world robots in an efficient and effective way. We showcase how the introduced modular toolbox can be applied to three different robotic tasks.

View Post

Guided Reinforcement Learning – Towards Efficient and Effective Real-World Robotics

July 12th, 2023

Julian

Eßer

Guided Reinforcement Learning describes a concept for learning control policies of intelligent robots. By incorporating additional knowledge, the training process becomes more data-efficient and effective for real-world robotics.

View Post

Introduction to Reinforcement Learning – A Robotics Perspective

July 5th, 2023

Julian

Eßer

Reinforcement Learning is a data-driven approach to learn intelligent behaviors through trial and error interaction with the environment. Related to robotics, it offers new chances for learning robot control under uncertainties for challenging robotic tasks.

View Post

Reinforcement Learning | LAMARR

Guided Reinforcement Learning – Application to three Robotic Tasks

Guided Reinforcement Learning – Towards Efficient and Effective Real-World Robotics

Introduction to Reinforcement Learning – A Robotics Perspective

Pipeline Step 1: Problem Representation (States, Rewards, Actions)

Pipeline Step 2: Learning Strategy (Algorithmic Design Decisions)

Pipeline Step 3: Task Structuring (Breaking Down the Problem)

Pipeline Step 4: Sim-to-Real (Bridging the Gap)

Summary