Reinis Cimurs

About Me

Hey there! I’m a machine learning research engineer with a background in Intelligent Robot Engineering. I’m really into autonomous navigation and decision making, and I’m passionate about finding ways to use A.I. in real-world scenarios.

I have hands-on experience with Deep Reinforcement Learning-based neural networks for navigation—taking sensor data and using it for robot navigation with ROS in real-life situations. As part of the automotive industry, I currently work on attention-based neural networks for driver modeling using real-world data and Graph Neural Networks.

My primary programming language is Python, but I also use C and Matlab when needed. I’m comfortable building deep learning models with TensorFlow and PyTorch.

Projects

Goal Driven Autonomous Exploration
A goal-driven autonomous exploration system that combines reactive and planned robot navigation in unknown environments.

[GitHub Repo]
DRL-robot-navigation
Deep Reinforcement Learning for mobile robot navigation in ROS Gazebo using Twin Delayed Deep Deterministic Policy Gradient.

[GitHub Repo]
DRL-robot-navigation-IR-SIM
Deep Reinforcement Learning for mobile robot navigation in IR-SIM simulation using DRL algorithms (SAC, TD3, PPO, DDPG, RNN).
[GitHub Repo]
Curvature Constrained Path Smoothing with Bezier Curves
Python methods for creating and optimizing quadratic and cubic Bezier curves with path smoothing.
[GitHub Repo]

Select Publications

Goal-Driven Autonomous Exploration Through Deep Reinforcement Learning – An autonomous navigation system for goal-driven exploration of unknown environments.
[Paper Link]
Analyzing Closed-loop Training Techniques for Realistic Traffic Agent Models in Autonomous Highway Driving Simulations – Analysis of training methods for realistic agent modeling.
[Paper Link]
Goal-oriented Obstacle Avoidance with Deep Reinforcement Learning in Continuous Action Space – Using deep RL from depth camera inputs for goal-oriented navigation.
[Paper Link]
Time-optimized 3D Path Smoothing with Kinematic Constraints – Generation of a kinematically constrained 3D trajectory from a path planner.
[Paper Link]
Bezier Curve-based Smoothing for Path Planner with Curvature Constraint – A path smoothing method for two-dimensional path planners with curvature constraints.
[Paper Link]
Information-based heuristics for learned goal-driven exploration and mapping – Heuristics method for environment exploration accounting for DRL constraints.
[Paper Link]

Patents

Method for Evaluating a Traffic Scene with Several Road Users – US20240346922A1: A method for evaluating a traffic scene with several road users.
[Patents Link]
Method for training a machine learning model – DE102023203277A1: A method for training a machine learning model for training data containing a plurality of training data elements.
[Patents Link]