Literature Database Entry

duma2025generalizing

Jakub Grzegorz Duma, "Generalizing Cyclists' Intention and Trajectory Prediction Using Explainable Deep Learning Models," Master's Thesis, Telecommunication Networks Group (TKN), TU Berlin (TUB), October 2025. (Advisor: Marie-Christin H. Oczko; Referees: Falko Dressler and Thomas Sikora)

Abstract

Urban intersections are the most conflict-prone places for cyclists. Anticipating whether a rider will turn left, go straight, or turn right - and forecasting the near-term path -matters for safety, cooperative automation, and infrastructure planning. This thesis studies scalable, explainable prediction of cyclist intent and short-horizon trajectories from crowd-sourced smartphone GNSS, with an explicit focus on generalization to unseen intersections. We build a deterministic, auditable pipeline that stabilizes raw traces, performs city-scale map matching with Open Source Routing Machine (OSRM), extracts and labels intersection visits, and trains two coupled learners: a calibrated gradient-boosted intent classifier and a road-relative LSTM-decoder whose forecasts are softly conditioned on the intent probabilities. Evaluation uses Leave-One-Intersection-Out (LOIO) protocols and reports geometric accuracy ADE/FDE, map compliance, and calibration-based reliability. Across identical test visits, the map-aware, two-stage design transfers better to held-out junction geometries than intent-agnostic decoders or raw-frame features: late-horizon displacement decreases, endpoints concentrate on the executed arm, miss rates fall on ambiguous (left/right) maneuvers, and uncertainty widens up-stream and contracts after commitment in a calibrated manner. Attribution analyses indicate that decisive cues accumulate mid-history in the ego/road-relative frame and that primary kinematics dominate. By decomposing behavior (intent - motion), representing geometry in road-relative coordinates, and calibrating predictive probabilities, the system delivers transferable and trustworthy forecasts from commodity data. The pipeline is reproducible and lightweight, providing a practical base for richer map encoders and interaction-aware probabilistic decoders.

Quick access

BibTeX BibTeX

Contact

Jakub Grzegorz Duma

BibTeX reference

@phdthesis{duma2025generalizing,
    author = {Duma, Jakub Grzegorz},
    title = {{Generalizing Cyclists' Intention and Trajectory Prediction Using Explainable Deep Learning Models}},
    advisor = {Oczko, Marie-Christin H.},
    institution = {Telecommunication Networks Group (TKN)},
    location = {Berlin, Germany},
    month = {10},
    referee = {Dressler, Falko and Sikora, Thomas},
    school = {TU Berlin (TUB)},
    type = {Master's Thesis},
    year = {2025},
   }
   
   

Copyright notice

Links to final or draft versions of papers are presented here to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted or distributed for commercial purposes without the explicit permission of the copyright holder.

The following applies to all papers listed above that have IEEE copyrights: Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

The following applies to all papers listed above that are in submission to IEEE conference/workshop proceedings or journals: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible.

The following applies to all papers listed above that have ACM copyrights: ACM COPYRIGHT NOTICE. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Publications Dept., ACM, Inc., fax +1 (212) 869-0481, or permissions@acm.org.

The following applies to all SpringerLink papers listed above that have Springer Science+Business Media copyrights: The original publication is available at www.springerlink.com.

This page was automatically generated using BibDB and bib2web.

Last modified: 2026-04-21