Contextual Observability of Software-Defined Vehicles

Software-defined vehicles (SDVs) address rising software complexity, reduce electronic control units (ECUs), and separate hardware from software, allowing for easier updates and enhanced vehicle lifecycle management. Over-the-air (OTA) updates offer dynamic functionality and improved user interaction, while SDVs provide cost efficiency, weight reduction, and faster time to market. However, challenges include achieving comprehensive observability in distributed architectures, cybersecurity risks, software maintenance complexities, high development costs, and data privacy concerns. This project aims to develop a testbed for SDV contextual observability. This testbed will enable collecting multimodal telemetry data, facilitating continuous monitoring, advanced analytics, causal inference and incident response to proactively detect and mitigate issues.
Funding agency: IRC for Smart Mobility and Logistics (SML) at KFUPM
Duration: 2025-2027
Topics: Software-defined vehicles, contextual observability, adavanced analytics, causal inference, automated incident response.

Agentic AI-based Framework for Seamless Integrated Mobility

Aligning with Saudi Vision 2030, this project supports Saudi Arabia’s goals to increase public transit use and improve accessibility for all citizens, including individuals with disabilities, the elderly, and low-income groups. The research focuses on developing an agentic AI-based framework for Seamless Integrated Mobility (SIM), envisioned as a unified platform that integrates multimodal transportation options—including private vehicles, micro-mobility solutions, ride-hailing services, and public transit—alongside supporting infrastructure such as charging stations, parking facilities, and toll roads.
Funding agency: KFUPM Deanship Research
Duration: 2025-2026
Topics: Seamless integrated mobility, agentic AI, end-to-end planning, service bundling, context-aware services, supply-demand matching.

General Motors Projects

Before joining KFUPM, the PI served as the AI and Smart Mobility Technical Leader at General Motors Canada. He led AI/ML projects focused on software-defined vehicles, connected and automated driving technologies, active safety systems, and prognostics, achieving successful technology insertions. He co-invented and filed 72 patents, trade secrets, and defensive publications, earning recognition as “Inventor of the Month” multiple times by GM's Global Patent & Invention Management team. The advanced features developed were for future GM vehicle models; therefore, further details cannot be disclosed due to confidentiality agreements with GM. Additional information about the filed patents can be found here.

VRU Crossing Intent Prediction

This research project introduces an innovative framework for pedestrian crossing intention prediction. The framework incorporates an image enhancement pipeline, which enables the detection and rectification of various defects that may arise during unfavorable weather conditions. Subsequently, a transformer-based network, featuring a self-attention mechanism, is employed to predict the crossing intentions of target pedestrians. This augmentation enhances the model's resilience and accuracy in classification tasks. Through evaluation on the JAAD dataset, our framework attains state-of-the-art performance while maintaining a notably low inference time. Moreover, a deployment environment is established to assess the real-time performance of the model. The results of this evaluation demonstrate that our approach exhibits the shortest model inference time and the lowest end-to-end prediction time, accounting for the processing duration of the selected inputs.
Funding agency: IoT Research Laboratory, Ontario Tech University
Duration: 2022-2024 | Topics: image enhancement, self-attention mechanism, vision transformers.

Robustness of Deep Learning-based VRU Detection Models

This research project highlights the critical role of accurate pedestrian detection in assisted and automated driving systems to enhance road safety. Real-world deployment faces challenges like image corruption and occlusions, addressed here through robust, stylized, and occluded training techniques. Robust Training uses intentionally corrupted examples to simulate real-world scenarios, significantly improving model resilience to various corruptions. Categorized training further refines performance by tailoring augmentations to specific corruption types. Stylized Training employs Adaptive Instance Normalization (AdaIN) to introduce texture and style variations, enriching the dataset. This approach boosts pedestrian detection accuracy, achieving notable improvements in mean Average Precision (mAP50), particularly in handling noise, blur, and weather-related distortions. Occluded Training generates datasets simulating different occlusion levels, enhancing model performance on occluded samples with a ∼2% increase in mAP50. This method demonstrates effectiveness in improving detection in challenging scenarios. Overall, the combined techniques improve model adaptability and robustness in diverse conditions, achieving a ∼2-4% performance boost. This work establishes a solid foundation for deploying reliable pedestrian detection models in complex environments, advancing the safety and reliability of automated driving systems.
Funding agency: Nile University
Duration: 2022-2024 | Topics: assisted and asutomated driving, VRU detection, model robustness

Optimal Placement of Bus Stops

Bus systems play an important role in the modern city. Carefully designed bus stop locations can lift up overall transportation efficiency and save more time for passengers. In this paper, Particle Swarm Optimization (PSO)-based approach is proposed to find the optimal placement of bus stops in Waterloo/Kitchener area. The selection of the bus stops placements takes into account neighborhood population, family income, age distribution, and other factors. The overall goal is to minimize the average travel time of the passengers which is composed of two parts namely passenger time in the bus and out of the bus. Experimental results on real bus lines in Waterloo/Kitchener region, Canada showed that both PSO and adaptive PSO provide shorter average commuting time compared to the original bus routes. Moreover, the optimized placement needs less number of bus stops compared to the original number of stations.
Funding agency: University of Toronto
Duration: 2022-2023 | Topics: optimal placement, swarm intelligence, particle swarm optimization.

Future of Public Transport Experience

E-Payment for public transport is a use case built by DM TECH featuring smart bus-station experience. The concept was designed to visualize the influence of digitalizing public transport accessibility, ticketing and payment. The experience is also available in VR for user- friendly and interactive simulation.
This work was conducted under direct supervision of the PI in his capacity as CTO of Disruptive Mobility Tech (DMTech).

Smart City Walkthrough

Smart city concept designed by DM TECH Experience Design team. The concept feature a street walkthrough integrating different technologies to support smarter, sustainable and personalized city experience. The experience is also available in VR for user- friendly and interactive simulation.
This work was conducted under direct supervision of the PI in his capacity as CTO of Disruptive Mobility Tech (DMTech).

Virtual ride experience for autonomous driving

A virtual ride experience for autonomous driving that Mobileye introduced a model-based approach to safety entitled Responsibility-Sensitive Safety (RSS) (Shalev-Shwartz et al., 2017). RSS highlights five safety rules an automated driving vehicle should be able to follow. These five rules include safe distance (i.e., the self-driving vehicle should not hit the vehicle in front), cutting in (the self-driving vehicle should be able to identify when lateral safety may be compromised by a driver unsafely crossing into its lane), right of way (the self-driving vehicle should be able to protect itself against human drivers who do not properly adhere to the right of way rules), limited visibility (be cautious in areas with limited visibility), and avoiding crashes (the self-driving vehicle should avoid a crash without causing another one). The environment, vehicle interior and assessment criteria were built to enable virtual testing and passenger-centric feedback collection using VR.
This work was conducted under direct supervision of the PI in his capacity as CTO of Disruptive Mobility Tech (DMTech).

Hyperloop Station Concept: El-Waha Revival

El-Waha (The Oasis) Hyperloop station concept simulate the contribution of transportation technology in building the future. The revival story features accessibility, availability, smartness, and design creativity to support smart city infrastructure and user expectations. Architecture concept, design, model, and visualization is owned by DM TECH. The experience is also available in VR for user- friendly and interactive simulation.
This work was conducted under direct supervision of the PI in his capacity as CTO of Disruptive Mobility Tech (DMTech).

Hyperloop Lab Facility

Here is a quick tour in the first Hyperloop Lab Facility in the world! The Hyperloop is a disruptive solution for the future of mobility and high speed transport. And despite the fact that the technology have been there for a long time now, there isn't still information accessibility and testing availability worldwide. This lab facility can be used for professional and educational training on Hyperloop technology and other related technologies. It allows students, trainees and researchers to study, experiment and understand different aspects of Hyperloop and similar technologies as disruptive transportation systems.
This work was conducted under direct supervision of the PI in his capacity as CTO of Disruptive Mobility Tech (DMTech).

Façade cleaning robot

Z21 is a smart automated façade cleaning system that encompasses a rooftop robot and a cleaning robot. The rooftop robot is a 2-degrees of freedom (DOF) motorized rooftop gantry crane that is responsible for positioning the cleaning robot and carries all the cleaning reagent tanks as well as the hanging cables and computation tools. The cleaning robot is equipped with advanced motion and stabilization mechanisms and can automatically inspect and clean glass windows and façades. It has much higher cleaning capacity compared to state-of-the-art cleaning systems and manual cleaning and is designed to work in severe weather conditions. This system is suited well for cleaning wide variety of façades, flat and disposed surfaces of high-rise buildings.
This work was conducted under direct supervision of the PI in his capacity as AI Devision Head at Sypron Solutions.

Agatha

Predictive maintenance is a cornerstone of Industry 4.0. Agatha is a predictive maintenance system via cognitive IoT. This system encompasses spatially distributed interoperable and accessible smart sensors able to selectively collect, fuse and share data about machine condition. The selectively collected and fused data is then analyzed in order to get real-time insights and performance data, determine and dynamically update the likelihood of failures and make timely decisions or recommendations. Agatha listens to the heart of a critical equipment, lets it communicate how it feels and offers advice whenever it needs looking after. Engineers will no longer need to manually gather data and generate reports for each critical equipment. Maintenance schedules can be easily planned by avoiding costly downtime. Productivity is increased, equipment lifetime is extended, energy is saved and maintenance cost is cut and unplanned stops are reduced or eliminated. Agatha can be used to monitor and track stationary and mobile assets in different smart city sectors such as smart governance, smart factory, smart energy, smart building, smart mobility, smart infrastructure, smart technology, smart healthcare and smart citizen.
This system was architected by the PI in his capacity as AI Devision Head at Sypron Solutions.

MineProbe

MineProbe is a minefield reconnaissance and mapping system that encompasses a number of spatially distributed unmanned ground vehicles (UGVs) equipped with an efficient multimodal landmine and unexploded ordnances (UXO) detection system and accurate hybrid localization system. The UGVs have been endowed with the ability to move fluidly and efficiently in rough terrain of North West Coast of Egypt. A centimeter-level accuracy outdoor hybrid localization system for UGVs is developed in this project. A GPR-EMI dual sensor for landmine detection with high detection rates, low false alarms and low false negatives. The system provides a mine map for the minefield that shows the exact locations of the detected landmines and UXOs.
This project was conducted under direct supervision of the PI in his capacity as Autonomous Vehicles Professor at Zewail City, Consultant at Innovision and PI of the project MineProbe.