Kumar Manas

PhD Student in Mathematics & Computer Science at Freie Universität Berlin

Specializing in Knowledge Formalization, Trajectory Prediction and Planning for Autonomous Agents

Kumar Manas

News & Announcements

New April 2024
Paper Accepted at RSS 2025!

Our paper "Uncertainty-Aware Trajectory Prediction via Rule-Regularized Heteroscedastic Deep Classification" has been accepted at Robotics: Science and Systems (RSS) 2025 conference.

About Me

I am enrolled in joint PhD program at Freie Universität Berlin and Continental Automotive, supervised by Prof. Adrian Paschke. I am specializing in the intersection of artificial intelligence and autonomous systems.

My research is dedicated to developing innovative methods for formalizing traffic rules and integrating them into autonomous driving systems. Additionally, I work as a research scientist at Continental Automotive, where I focus on translating these theoretical frameworks into practical applications.

Current Research

Trajectory Prediction with Rules and Uncertainty

Working towards integrating traffic rules into the trajectory prediction module in a implicit manner. Alongside we are improving uncertainty quantification in the prediction.

Trajectory Planning with Rule Integration

Algorithms for planning vehicle trajectories in complex traffic scenarios while adherring to the rules of the road.

Publications

Interactive Demonstration: Traffic Rule Formalization
Uncertainty-Aware Trajectory Prediction via Rule-Regularized Heteroscedastic Deep Classification

Proceedings of the Robotics: Science and Systems (RSS), 2025

Kumar Manas, Christian Schlauch, Adrian Paschke, Christian Wirth, and Nadja Klein

Paper | Project Website |

CoT-TL: Low-Resource Temporal Knowledge Representation of Planning Instructions Using Chain-of-Thought Reasoning

2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

Kumar Manas, Stefan Zwicklbauer, and Adrian Paschke

Poster | Presentation

Optimierung der Entscheidungsfindung in autonomen Fahrsystemen mit neuro-symbolischem Wissen

Book Title: Hybride KI mit Machine Learning und Knowledge Graphs

Ya Wang*, Kumar Manas*, and Adrian Paschkee

TR2MTL: LLM-based Framework for Metric Temporal Logic Formalization of Traffic Rules

IEEE Intelligent Vehicles Symposium (2024)

Kumar Manas, Stefan Zwicklbauer, and Adrian Paschke

Poster |

Legal Compliance Checking of Autonomous Driving with Formalized Traffic Rule Exceptions

Workshop on Logic Programming and Legal Reasoning in International Conference on Logic Programming (ICLP) (2023)

Kumar Manas and Adrian Paschke

Semantic Role Assisted Natural Language Rule Formalization for Intelligent Vehicle

International Joint Conference on Rules and Reasoning (RuleML+RR) (2023)

Kumar Manas and Adrian Paschke

Presentation

Patents

Contact