Interview with Samuel B Siewert, Software Engineer, Embry Riddle Aeronautical University - Speaker at 5th Annual Global Big Data Posted on : Jul 25 - 2017

We feature speakers at 5th Annual Global Big Data Conference - August 2017 to catch up and find out what he or she is working on now and what's coming next. This week we're talking to Samuel B Siewert, Software Engineer, Embry Riddle Aeronautical University (Topic : Drone Net - Multi-Modal Sensor Network For SUAS Shared Air Space Safety And Security)

Interview with  Samuel B Siewert

1. Tell us about yourself and your background.
I’ve spent half my industry career working on NASA astronautics and deep-space projects (Houston, JPL AI group, and Spitzer space telescope) and the other half on commercial products for storage and networking ranging from chip-level firmware for fiber channel, SAS (Serial Attached SCSI) up to scalable SAN/NAS RAID solutions for high-throughput computing.  I left industry in 2012 to teach and research full-time and presently have a tenure-track position in Computer, Electrical, Software Engineering at Embry Riddle and an adjunct appointment to a professional Master’s program and the graduate school at CU Boulder.

2.  What have you been working on recently?
Drone Net – a joint research project involving Embry Riddle and U. of Colorado Boulder.  The concept is to develop an open systems reference design for small UAS air traffic management, security and safety in urban, campus, and other challenging and air-traffic sensitive environments.
We are working on defining new algorithms for data analytics (machine vision and machine learning), but first build out of a test system with a network of multi-sensor, multi-modal nodes that we want to cost no more than a workstation or server would.

3. Tell me about the right tool you used recently to solve customer problem?
Our customers are students and research sponsors.  For students, I have found the NVIDIA stack for embedded systems, JetPack 3.0 on Tegra quite useful and use it for teaching in my ECEN 5763, EMVIA course at CU Boulder.  Likewise I have found the Linux Foundation Zephyr open software useful for IoT and embedded systems along with traditional RTOS and embedded Linux, which I use for teaching CEC450, Real-Time Systems at Embry Riddle.  Most of what I use to teach is likewise useful for sponsored research.  As we dig deeper into our research we are finding many open source AI and machine learning tools useful, including: Tensorflow and CUDA of course.  We are developing our machine learning on a Lambda DevBox and an Intel Xeon Phi workstation, comparing many-core to GP-GPU acceleration.  Overall, we try to be brand agnostic and keep our solutions open.

3b. Where are we now today in terms of the state of artificial intelligence, and where do you think we’ll go over the next five years?
Wide spread commercial use of AI is starting to really take off with many practical and significant solutions for challenging problems in transportation, data mining, science and engineering.  AI has been stuck in the lab or limited to specific domains in the past as well as suffering a cycle of over-exuberance followed by disappointment – at this point it appears to have hit a more stable productive phase of evolution based on finding tractable problems with clear objectives for success (NLP assistants, self-driving cars, recommendation engines, etc.).  So, AI has matured to a more product and service oriented stage, but there’s still plenty of basic research to be done as well, so an exciting time to work on applied and basic AI R&D.

4. Where are we now today in terms of the Big data, and where do you think we’ll go over the next five years?
High throughput computing has become very significant as workloads of interest have started to shift from and balance out with simulation and processing heavy workloads to data driven workloads.  The other driver of Big Data is IoT (Internet of Things - sensor networks) and availability of mobile and field use data generating instruments.  The human visual system includes essentially two 50 megapixel (rods + cones) per eye intelligent system and today each and every car will soon have the equivalent.  Most of this data can be generated and consumed on an embedded system, but Cloud data analytics for intelligent transportation and other sensor networks can greatly enhance safety and reliability of intelligent systems if they can be made secure, scalable, with good wireless access.  While raw data used to be the problem, now even detection or identification operations per second can overwhelm.  Big data will have a role in mobile intelligent systems much like it does for mobile smart phones and tablets, but now with vehicles, UAS, security, medical, and many new or revitalized industries.

5. You’ve already hired Y number of  people approximately. What would be your pitch to folks out there to join your Organization? Why does your organization matter in the world?
We are not for profit, applied research focus, but with a goal to enable and tinker with basic research challenges in machine vision and learning. Many upstart and fortune 500 for profit companies and government agencies are developing small UAS ATC solutions for safe/secure/compliant operations [1) Blacksage, 2) Droneshield, 3) Dedrone, 4) Gryphon, 5) AARONIA, 6) UTM, 7) LATAS to name a few].  Droneii does a good job of tracking and characterizing this market.  Much like general aviation faced in the 1920’s and 30’s, we have a huge growth in small UAS applications and use, and it’s a market the US can’t afford to ignore or tie up with regulations that over limit, yet small UAS must share airspace with GA, share proximity with people and buildings, and must be safe.  The FAA has created a university research group to address this – FAA ASSURE.  Embry Riddle has been a founding university participant in this program and recognizes the significance, opportunity and risk with a unique historical perspective as well as leading edge research in aviation and related engineering disciplines.  It’s a nature fit for Embry Riddle to assist with safe, secure, scalable integration of small UAS with national airspace.

6. What are some of the best take aways that the attendees can have from your workshop on "Drone Net - Multi-Modal Sensor Network For SUAS Shared Air Space Safety And Security"?
This is a Big Data problem.  Collecting data from small UAS and methods to track, identify and localize small UAS are not lacking, but efficient, effective and scalable protocols and algorithms to deal with both compliant and non-compliant small UAS do need attention.  Development with existing knowledge, applied research to characterize performance and the problem space as well as some basic research are all needed.  Solutions from the military and NASA will facing scaling and cost challenges as well, so this needs to be a joint effort between industry, university researchers, government agencies, FFRDCs and the public (to be compliant).  Simple tools to know what’s in the air (similar to flightradar24.com for example) are needed for small UAS.  Universities are uniquely positioned to provide testing, research, and to help train the future generations of UAS operators, engineers, and policy makers.

 
7. What are the top 5 Big data Use cases in enterprises?

1. Intelligent transportation (self-driving cars, aircraft, trucks and trains)
2. Security and safety systems (physical)
3. Cybersecurity
4. Medical and health care
5. IoT, citizen science, and public sentiment analysis

8. Which company do you think is winning the global Big Data race?
It’s hard to pick one because Big Data requires layered architectures (hardware, firmware, software, human interface) and based upon the need for a Big Data ecosystem.  Fundamental architecture that enables Big Data also comes from embedded/mobile systems as well as data centers and high throughput computing.  Since I can’t pick just one, I will rank my top 10 in terms of my opinion on significance and impact (including a few non-profit): 1) Google (wide use AI), NVIDIA (GPU AI architecture revolution), Intel (Data centers), Tesla (early Autopilot), IBM (Watson), Apple (HCI), iRobot (home robotics), Xilinx (FPGA efficiency), Linux Foundation (open source and Zephyr), and DJI (simple to fly drones for masses). 

9. Any closing remarks
I think more partnerships between industry, government and academia will assist not only with Big Data and AI development, but with public acceptance and adjustment.  We need to educate a whole new workforce in the new millennium, create new policy, and integrate AI into embedded and scalable systems in a way that enhances quality of life without a real or perceived threat to individual security, safety and financial well-being.  Recall that in the late 1930’s the panic in the population related to science and aviation was so severe that a radio broadcast by Orson Welles of H.G. Well’s “War of the Worlds” lead to wide spread panic.  We have a great challenge, risk and opportunity presented with AI and Big Data.