30hp-Feasibility Of Visual Models For Real-Time Explainability In Surround-View Autonomous Driving

Introduction: Thesis work is an excellent way to get closer to Scania and build relationships for the future. Many of today's employees began their Scania career with their degree project.

**Background: **Autonomous vehicles rely on complex perception and planning pipelines that are often opaque. For safe deployment, systems must not only act but also explain their reasoning in human-understandable terms. Recent Visual-Language Models (VLMs) show promise in generating natural-language descriptions of visual scenes, yet their feasibility for real-time, on-vehicle explainability remains unexplored—especially in surround-view settings where multiple cameras capture a 360° environment.

**Problem: **This thesis investigates whether VLMs can generate trustworthy, real-time explanations of driving decisions under the latency and resource constraints of automotive hardware, while handling multi-camera inputs efficiently.

Research Questions:

• Can VLMs provide natural-language explanations within strict real-time budgets (<100 ms)?
• Do the explanations align with actual driving events and human expectations?
• How can surround-view inputs be processed for VLMs without exceeding compute limits?
• How robust are the explanations under adverse or out-of-distribution conditions?

Objectives:

• Benchmark state-of-the-art VLMs for latency and throughput on GPU and embedded platforms.
• Develop a pipeline for surround-view fusion and efficient input handling.
• Propose methods to ground explanations in structured driving representations (lanes, maneuvers, traffic rules).
• Evaluate explanation faithfulness, clarity, and safety relevance through automatic and human studies.

Education/program/focus:

Indicate education, program or focus: Masters program on computer science with a focus on AI

Number of students: 1 Start date for the thesis work: January 2026 Estimated time required: 6 months

Contact persons and supervisors: Mohammad Nazari, Ph.D., Mohammad.nazari@scania.com

Application: Your application must include a CV, personal letter and transcript of grades

A background check might be conducted for this position. We are conducting interviews continuously and may close the recruitment earlier than the date specified.