Dr. Yazan A Alqudah

In this paper a model-based embedded systems design laboratory course development is presented. The course is an enhancement over an existing embedded systems laboratory course that uses conventional methods along with low-level and high-level programming languages in designing embedded systems. The proposed laboratory course introduces the concepts of model-based design, rapid prototyping, and auto-code generation to junior students of an undergraduate electrical engineering program. The students are exposed to model-based development and design through using MATLAB ® Simulink ® . The graphical environment of Simulink ® allows students to easily design and implement embedded software and generate code without worrying about the details of conventional coding issues. Moreover, MATLAB ® Simulink ® allows easy auto-code generation to be executed on a variety of microcontrollers and FPGAs. The developed lab course and its experiments were implemented for two semesters at least. One of them was during the COVID-19 shutdown and thus the lab was conducted by students at home. The students' feedback is promising and shows that the lab has helped attain more skills needed by the industry in addition to acquiring a new knowledge area that was not covered by the conventional curriculum.

In this work, a low-cost device is developed that enables the monitoring and classification of sleep positions. Driven by the growing problem of sleep disorders, the device can be utilized at the patients’ place to record and report their sleep positions. The device can also be used by hospitals and clinics for patients that requires continuous monitoring. Machine learning is used to classify different sleep positions based on sensors data collected from the device.

The reliability of autonomous vehicles can be enhanced by providing the vehicle with more information about the surrounding environment. Autonomous vehicles typically use LiDAR, Radar, and computer vision to substitute for the driver’s vision. By adding auditory perception, an autonomous vehicle will improve its reliability and enable the vehicle to react better to the environment.
This paper proposes a novel approach to enhance the reliability of autonomous vehicles. By adding auditory perception, the vehicle will be able to hear and process audio. To illustrate the advantage of auditory capability, we conducted an experiment to collect and process audio signals obtained from a vehicle driving on the road. We showed how an audio signal can be processed to obtain extra information that can alert the vehicle to potential dangers.

Motion estimation with a quad-tree variable block size is the driver for the high performance of HEVC in video compression. However, tremendous sum of absolute difference (SAD) computations are needed for every single block size, resulting in huge memory access and power consumption. This work proposes an early termination mechanism at the hardware level that is suitable for quad-tree variable block size motion estimation without degrading video quality. The simulations show reductions in memory access to fetch pixels of about 34.03% and 51.98% saving of SAD computations. This reduction results in reduced power consumption. The simplicity of the algorithm makes it a hardware friendly mechanism, which could be adopted in any motion estimation hardware design.