MS Thesis - High-Fidelity Motorcycle Simulator for Rider Behavior Research

Background

Motorcycle safety research requires detailed and repeatable measurements of both rider behavior and vehicle dynamics, particularly during demanding or safety-critical maneuvers.

A key challenge is understanding rider behavior immediately before hazardous events, as such knowledge is critical for developing future rider-assistance and safety systems. Real road testing can provide valuable information, but it is constrained by safety, repeatability, scenario control and data observability.

An existing motorcycle riding simulator integrates a physical riding platform, BikeSim vehicle dynamics, CARLA and Unreal Engine 5 visualization, a Python supervisory application, a C/C++ integration layer, sensors, steering force feedback and motion-platform feedback. The basic closed-loop integration has been established. The next step is to improve physical consistency, rider-perceived steering fidelity, vehicle response and end-to-end system performance, while creating a reliable experimental data-acquisition platform.

The project will therefore focus on improving and validating the simulator as a human-in-the-loop research platform. The intended result is a practical system that balances simulation realism, controllability, user experience, stability and experimental repeatability, while collecting synchronized high-resolution data describing both the motorcycle and the rider.

Aim

The aim of this thesis is to improve and validate an existing motorcycle riding simulator as a high-fidelity experimental platform for rider-behavior and pre-accident safety research. The work will investigate the relationship between simulator fidelity and rider perception, with particular focus on rider trust, controllability, and behavioral realism. The resulting platform should support synchronized acquisition of rider movement, steering input, interaction forces, vehicle dynamics, and virtual-environment data.

Research Questions

• How can steering force feedback and vehicle response be improved to provide coherent and realistic rider cues?
• How can high-resolution rider and vehicle measurements be synchronized and recorded reliably across heterogeneous real-time systems?
• How do simulator characteristics influence experienced riders' perception of realism, controllability and trust in the simulator response?

Technical Platform

• Physical motorcycle riding and motion platform
• BikeSim motorcycle dynamics model
• CARLA/UE5 simulation environment
• Python-based control, supervision and logging software
• C/C++ DLL and simulator integration components
• Steering servo and torque measurement
• Wearable IMUs, fixed cameras and supporting sensor infrastructure

Project Objectives

• Review and extend the current end-to-end signal chain between the physical platform, Python, CARLA/UE5, the C/C++ integration layer and BikeSim.
• Improve real-time data exchange, observability, logging, timing and fault handling.
• Provide riders with a realistic and coherent riding experience while maintaining physical consistency.
• Improve vehicle-control behavior, including low-speed stability, steering-input mapping, lean response, corner entry and recovery to upright riding.
• Evaluate the consistency between rider input, BikeSim vehicle response, visual feedback and physical feedback.
• Design and implement scenarios in CARLA that include road layouts, environments, traffic participants and events.
• Validate the correctness of system operation and feedback across a range of scenarios.
• Use fixed-camera video for experiment documentation and review. Camera-based automatic pose estimation may be included as an optional extension.

Study Design

The development and verification process will be iterative. Issues identified during verification or validation will be corrected and tested again. The rider experiment will begin only after the technical verification and scenario-based validation have been completed successfully.

The study is divided into five main steps:

• System FamiliarizationThe student will become familiar with the existing simulator, including its components, data flow and operating workflow.

• System ImprovementThe student will improve workflow and data exchange between Python, the C/C++ DLL, BikeSim and CARLA/UE5. The work will also improve steering-input mapping, vehicle-control behavior and steering force feedback, including resistance, damping and increased resistance at large steering angles.
• Component and Integration VerificationThe student will verify that control functions and data transmission work correctly within each component and between components. Timing, latency and synchronization will also be evaluated.
• Scenario-Based ValidationThe complete simulator will be validated in selected riding scenarios to confirm that vehicle behavior, control responses and system feedback are correct, consistent and repeatable.
• Rider ExperimentExperienced motorcycle riders will test the simulator in selected scenarios. Objective measurements and rider feedback will be used to evaluate realism, controllability and the overall riding experience. The experiment will follow applicable requirements for participant safety, informed consent, data protection and ethical approval.

Expected Deliverables

• An improved and operational motorcycle riding simulator suitable for the selected experiments.
• Implemented software improvements for real-time integration, steering/vehicle response and data acquisition.
• A documented, synchronized data-recording pipeline for rider, vehicle, platform and virtual-environment data.
• A defined set of test scenarios and experimental protocols.
• A well-documented codebase, configuration and deployment guide.
• Documented results from component verification, scenario-based validation and rider evaluation.
• Final thesis report, recommendations and an internal presentation.

Suitability
1-2 master's thesis students with an engineering background.

Required :

• Python and/or C/C++

Meritorious :

• Experience of CARLA or comparable simulation environments
• Control systems and vehicle dynamics
• Signal processing and sensor integration
• Experimental design and quantitative data analysis
• Benefits and Learning Opportunities

• Human-in-the-loop simulation
• Vehicle dynamics using BikeSim
• CARLA and Unreal Engine 5
• Real-time control systems
• Motion platforms and force feedback
• Research methodology and experimental validation

Industry-relevant motorcycle safety research

Contact

Christian-Nils Boda

christian-nils.boda@autoliv.com

More lives saved - more life lived!