Paul Karam

Paul Karam

Director of Engineering, Quanser 

Talk Title: Focus on Research, not Low-Level Tasks: Open Architecture Design Environment for Unmanned Systems

 

 

 

 

 

 

Biography:

As Quanser’s Director of Engineering, Paul Karam has spent the last 10 years leading the engineering team in its mandate to develop transformational labs that empower a global community of educators and researchers. Paul’s vision and guidance is essential to ensuring that the R&D team is able to deliver solutions that are effective, sustainable, and academically appropriate.

Paul plays a key role in the agile development process which focuses on assisting our partners attain their educational, research and development goals. His achievements in this area include the Rehabilitation Robot developed for the Toronto Rehabilitation Institute and the Quanser Autonomous Vehicle Research Studio.

Before assuming his current position, Karam managed Quanser’s industrial projects and robotics technologies for six years. He was involved in the design and implementation of Quanser’s original integration with National Instruments (NI) LabVIEW™ control design software and played a pivotal role in the development of the QNET line of engineering trainers for the NI ELVIS platform. He holds an honours bachelor degree in Electrical Engineering from the University of Waterloo.

Si-based millimeter-wave transceivers and associated antenna-in-package integration techniques have recently reached a high level of complexity and maturity. These advances have enabled the commercial deployment of millimeter-wave systems in mobile communication networks and automotive radar systems.

The complexity and reconfigurability enabled by the co-integration of digital and millimeter-wave circuits has also resulted in a practically infinite configurability space, especially in the context of beamforming systems. Current millimeter-wave beamforming systems employing hand-crafted configuration algorithms can access <1% of this space. At IBM Research, we envision a future where the application space of millimeter-wave systems enters new domains as the capabilities of such systems are fully realized by AI-driven adaptation and data interpretation. Effective mmWave hardware-to-software vertical integration is a key step towards the realization of this vision by closing the gap between antennas and AI.

This talk presents two examples of such vertically integrated systems. First, a Software Defined Phased Array Radio (SDPAR) is discussed. The SDPAR consists of a highly configurable 64-element dual polarized phased array operating at 28 GHz, a Software Defined Radio (SDR), and a software stack enabling access to both radio and spatial beamforming functions from a single API. The second example is a multi-spectral imaging platform consisting of a 60-GHz imaging radar, an IR camera, a visible domain camera, and software for the joint visualization of information from these three distinct spectral domains. Experimental results obtained with these platforms will be presented along with an outlook of their evolution towards the development of AI technology tailored to adaptive communication systems and portable imaging devices.

 

Abstract:

Unmanned systems research is often labor-intensive and time-consuming unless founded upon sound hardware, and a robust computation and communication infrastructure. Imagine, in addition, a direct access to low-level IO, all the way up to high-level application-focused commands, enabling researchers to rapidly build any application. Such an open architecture design environment would allow seamless deployment of research algorithms on day one. Additionally, every constant, gain or parameter would be customizable as required, all in the graphical and elegant Simulink environment. Equipped with such a system, researchers could dive right into applications such as visual servoing, multi-agent formation and path planning, swarms, etc. Find out how Quanser’s Autonomous Vehicles Research Studio empowers researchers to do this today along with cross-compilation for a variety of targets, simple communication setup, real-time and rapid control prototyping and more.