Raycasting Engine & SLAM Simulation

Raycasting Engine & SLAM Simulation Sandbox — Personal / Research Project

Coded a full ray-casting 3-D engine from scratch in 4 weeks to generate synthetic worlds for robotics SLAM research.

Date: Feb 2014 – Mar 2014
Roles: Solo Developer · Graphics Programmer · Research Assistant
Team size: 1 (mentored by two PhD supervisors)
Context: Frankfurt University of Applied Sciences — Robotics Lab
Core Stack:
Engine: C99 · SDL 1.2 · Custom Math Library
Tooling: GCC · gdb · Make
Extras: Qt-based Level Editor · Shooter Demo

Show more ↓

1. Problem

The robotics group needed varied, controllable 3-D environments to benchmark SLAM algorithms without expensive real-world data collection. Existing simulators were heavyweight, closed-source, or unsuitable for ray-sensor research.

2. Solution Overview

I built a minimal, real-time raycasting engine (inspired by Wolfenstein 3D) that exposes ground-truth camera poses for algorithm evaluation.

Module	Purpose	Key Tech & Concepts
Renderer	60 FPS column-wise raycaster with textured walls & sprites	DDA traversal · fixed-point arithmetic
Physics	Grid-aligned collision & movement	2-D vector maths
SLAM API	Step-through playback & pose export (CSV / UDP)	C socket API
Level Editor	Drag-&-drop map authoring	Qt Widgets · JSON

3. Impact

Produced 100 + synthetic mazes, accelerating SLAM prototyping by ≈3 × vs. real-lab captures.
Adopted in two MSc theses; cited in an internal research note.
Personal outcome: deep mastery of linear algebra, trigonometry and low-level optimisation.

4. My Contributions

Designed architecture; implemented all engine subsystems; built tooling (editor & shooter demo); integrated data export into research pipeline; documented maths & API for peers.

5. Key Challenges & Mitigations

Challenge	Mitigation & Result
Real-time speed on 2010-era laptops	Fixed-point math & cache-friendly DDA → stable 60 FPS
Accurate ground-truth pose export	Decoupled render loop from camera state → frame-exact logs
User-friendly map creation	Visual editor + JSON interchange → < 5 min scene setup