ppo-algorithm

AI-powered production line optimization using reinforcement learning (PPO).

This Legal Document Analyzer is a proof-of-concept NLP project demonstrating the potential of transformers for legal document summarization.

Stable Baselines3

Vision-free robotic pick-and-place using a 7-DOF Franka Panda in MuJoCo, combining PPO for high-level joint-position control with PID for low-level torque control. Includes curriculum learning, reward shaping, randomized object placement, and comparison with IK+PID.

Multi-modal RL trading agent (CNN + PPO) integrating market prices, macroeconomic indicators, and news signals . MSc dissertation artefact.

End-to-end safe robot navigation in crowds using ROS2. Chains LiDAR-based pedestrian detection (DR-SPAAM), SORT tracking, Gumbel Social Transformer trajectory prediction with conformal uncertainty bounds, and a PPO-Lagrangian constrained RL policy for safety-aware velocity control.

A specialized Reinforcement Learning (RL) project focused on multi-task mastery across 10 distinct gaming environments. General-Gamer-AI-Lite implements a lightweight multi-task agent designed to learn shared representations and transfer knowledge between varied game mechanics, from classic arcade challenges to strategic grid worlds.

ClimatePredictor implemented by using Proximal policy optimization (PPO) with ray framework for the FederatedLearning approach

Среда робота-уборщика на базе Gymnasium с обучением через s-b3 PPO

Autonomous Orbital Rendezvous & Docking using PPO + High-Fidelity Clohessy-Wiltshire Dynamics with Realistic Perturbations (NRLMSISE-00 Drag, J2, SRP, Thruster & Sensor Noise)

AI-driven robotic arm simulator trained with PPO Reinforcement Learning, featuring a real-time web telemetry dashboard.

AI-based Autonomous Driving Simulation using Reinforcement Learning. Implements a hybrid PPO + Imitation Learning architecture for efficient, safe policy optimization in MetaDrive.

12-DOF Unitree Go1 quadruped locomotion via custom PPO in MuJoCo — trained 6M+ timesteps to stable trotting at 0.6 m/s from scratch

AI agents for Trackmania using the TMRL package. Implemented DDPG, PPO, and used two SAC algorithms (with one or two critics) to train cars to navigate custom-built tracks.

Gridworld problem with dynamic obstacles implemented in PyGame GUI and solved using traditional RL agent (Q-learning) and deep RL agent (PPO algorithm).

AI - PPO-agent med CNN-policy for visuelle observasjoner. Agenten spiller mot en easy(eller annen difficulty) starcraft 2 bot i hoved filen som brukes (les readme) Masked actions sørger for at agenten velger valide handlinger av de 572 handlingene som er tilgjengelige.

Working on new variant of PPO implemented in predefined jvrc robot simulation and SLM model integration

RL agent that packs 3d boxes into a container

RL-based autonomous controller for a warehouse delivery robot navigating a constrained POMDP environment using sparse sensor data. CS780 Capstone Project.