Amin Assadi

SENSI (sensiperception.org): built from method to running system, quantifying how far an automated vehicle can trust its own perception. Probabilistic models of perception error, validated ground truth, LLM/VLM environment labelling, and an evaluation framework for camera-lidar-radar perception, extended into traffic simulation and motion planning. Lecturer and examiner (SSY300); supervised two MSc theses in applied AI and MSc project teams. Side project: AIEA, an open-source AI exam assistant.

Level 4 truck safety fallback system (for Continental), within an ASPICE process. Specified collision avoidance with vehicles, VRUs, animals and objects; co-specified cruising-in-lane behaviour. Owned functional and performance requirements; produced black-box and white-box design for perception, planning and interpretation. Carried a three-zone VRU concept to release under ISO 26262 and SOTIF. Built simulation, crash-analysis and false-braking tools. Mentored engineers; supported RFQs and PoCs.

PhD project, Safety in Mixed Traffic Environment (funded by AVL): showed that ADAS design should account for the risk the ego vehicle imposes on surrounding traffic. Double-layer stochastic driver model; qLPV models predicting the following driver's reaction, used inside the controller; MPC merging assistant and cautious highway pilot. Also built a stochastic microscopic traffic simulator with scenario generator, and drone-based traffic measurement.

Internship on bCUBE, a portable real-time PCR device for DNA and RNA detection. Built a MATLAB GUI toolbox for selecting the device's optical filters, with data-processing scripts and simulations to validate the selection.

Thesis: “Safety in Mixed Traffic Environment” Topics: Autonomous Driving, Prediction, Vehicle Control, Safety, Driving Behavior Model, Microscopic Simulation

Thesis: “Modeling, Control, and Energy Optimization of a Multi-Functional Building” Topics: Mixed Logical Dynamical System, Hybrid Model Predictive Controller, Energy Consumption Optimization