We have been talking about drones and related issues for a while now. This piece discusses recent news published on the MIT web site.
Drone researchers have been able to miniaturize every component of a drone but come against a wall when it comes to miniaturizing computer chips (brains of the drone). I knew the drones were shrinking but I never really thought about the limitations on how small they could get due to standard computer chips built for larger devices. Using a standard off the shelf computer chip in a drone – miniature or not – has to hit a size limit.
Most standard computer chips are energy hungry (page 3, Resource Utilization). This does not matter much if the device is a laptop or something else that has substantial power. However, for a small battery powered drone, a chip that eats up 10 to 30 watts is a significant drag on battery life. The smaller the drone, the bigger the issue.
Image and Control Management
Since many drones transmit still imagery and motion video, drone chips need to do a lot of image processing. As any computer gamer knows, good resolution images mean more memory and more power consumption. Besides image processing and streaming, drone controls and autopilot systems are a significant consumer too. Any energy that can be saved would go directly into a greater time of flight.
Integrated Design is the Key
The MIT team found that image processing and drone control algorithms that were written independently and then later implemented on the computer chip had issues. This traditional development method where the code and the hardware were designed separately and then married together was very inefficient, but they soon discovered a better way.
The team found if they developed the code and the chip together much greater efficiencies can be obtained. This innovation is less complicated than it sounds because of a chip they used called a field programmable gate array (FPGA). The FPGA is a simple chip that has tens of thousands of disconnected gates that can be connected in any order the developer desires. This approach allows the coders to build a computer chip that is precisely tailored to perform the functions they need (and does nothing else). The result is a built to order computer chip that uses less than 2 Watts of power. The research team thinks it can get this down to milliwatts. FPGA are also used extensively in self driving cars – notice the parallel with the automated drone.
Possible Military Applications?
Guess who is interested in this? The MIT research was partially funded by the Air Force.
David King Chief Technical Officer
Serving as the Chief Technology Officer, David is responsible for the discovery and implementation of new technologies that yield competitive advantages while working closely with Executive Management to develop strategies to increase revenue and performs cost-benefit and return-on-investment analysis.
In October 2005, David joined AEgis Technologies as Director of the Simulation Development Group. During that time, David helped formulate the direction of AEgis’ software development efforts to ensure maximum reusability, thereby speeding ongoing development while reducing costs. His focus is technology transfer to commercial products and product development. Under his leadership, the Simulation Development Group developed and fielded over 8000 UAS training systems across the DOD. David has held the title of Vice President, Simulation Development and most recently, Executive Vice President, Technology Solutions Division.
With more than 25 years in the Modeling, Simulation, and Training industry – and having served in product design and development, technology advancement, project management, and leadership positions. David is a recognized leader and innovator in the MS&T industry, having co-founded, matured and sold a two-man start-up company into a highly-successful small business (CG2, Inc.) in just seven years with no outside venture capital or investments. In addition to being recognized locally as an “Outstanding Small Business” for three consecutive years by the Madison County Chamber of Commerce.
Before co-founding CG2, David spent eight years in the US Army in missile-related fields. He worked for Electronic Associates Inc. and later AMTEC Corporation at Redstone Arsenal, Huntsville, AL, on a broad range of projects including the Sensor Vision III Real-Time IR Scene Generator, JAVELIN, STINGER, CHAPARRAL BAT, AIT, and THAAD HWIL simulations.
Specialties: Modeling, Simulation and Training. Hardware in the Loop, real-time simulations.