REAL-TIME OPTICAL FLOW SENSOR DESIGN AND ITS APPLICATION ON OBSTACLE DETECTION FOR UAVS
General Information
Motion is one of the most important features describing an image sequence. Motion estimation has been widely applied in structure from motion, vision-based navigation and many other fields. However, real-time motion estimation remains a challenge because of its high computational expense. The traditional CPU-based scheme cannot satisfy the power, size and computation requirements in many applications.
With the availability of new parallel architectures such as FPGAs and GPUs, applying these new technologies to computer vision tasks such as motion estimation has been an active research field in recent years. In this dissertation, FPGAs have been applied to real-time motion estimation for their outstanding properties in computation power, size, power consumption and reconfigurability.
This technology allows UAVs to automatically detect obstacles in flight on a real time basis without ground user direction. UAVs can be programmed with a specific flight path and then can adjust the flight path according to obstacles that it finds along that path and still end up at the desired location.
The Market
UAS Industry Background – (Unmanned Aircraft Systems)
At the present time, the primary market for UAS is the military. For the financial years 2007 to 2012, the US DoD has budgeted $7 billion for research, development, test, and evaluation (RDTE), $11.3 billion for procurement, and $3.5 billion for operation and maintenance of unmanned aircraft systems. These expenditures include UAS of all sizes. The figures above are dominated by the larger UAS, such as Global Hawk and Predator. Teal Group’s 2009 market study estimates that UAS expenditures will double within a decade from $4.4 billion annually to $8.7 billion, with more than $62 billion spent over the next ten years.
The study indicates that the U.S. will account for 72 percent of the worldwide RDTE spending on UAV technology over the next decade, and about 61 percent of the procurement. For small UAS (under 10 lb), G2 Solutions has forecasted the total revenues in excess of $540 million for years 2009 to 2012. It is expected that the military market will be the dominant outlet for UAS technology for the next 10 to 20 years.
Commercial civil markets are not well established and their development will require the definition of regulations for legal operation in the NAS along with the creation of new technologies to meet them. Markets will exist for both a service and product-based offering. There will be many companies that will want to contract surveillance or sensing capabilities at their facilities. The Forest Service, Division of Wildlife Resources, FEMA, and other disaster recovery and assessment companies will lean toward a service model in the early stages. Once the technology matures, the trend will be toward direct ownership and operation of unmanned aircraft systems. Currently, non-military expenditures are only a small fraction of the total spent on UAV technology.
We believe that non-military applications represent a significant complimentary growth area for UAS technology in the short and long term. While the potential for growth in non-military markets is difficult to quantify, we have made a significant effort (with the support of the State of Utah) to quantify the market opportunity for civil and commercial applications. Our findings have been eye opening. Aerial surveillance currently performed by manned helicopters and aircraft represents an application where unmanned aerial vehicles could have some impact.
Our initial research indicates that U.S. cities with populations over 250,000 spend approximately $250 million annually on helicopter patrol and surveillance. Our estimates also indicate that approximately $2.8 billion is spent annually on aerial inspection and monitoring of U.S. pipelines. While only information from pipeline monitoring is presented, other significant infrastructure monitoring applications exist.
Taking the current level of expenditures and projecting that 20 percent of manned aircraft surveillance hours could be replaced by small UAS surveillance results in a significant market opportunity. If we assume that the useful life of a UAS is 200 flight hours and that each unit includes one UA backup, approximately 10,000 units per year would be required to meet the market demand. If the cost of one unit is assumed to be $15K, the potential market is $150M per year for the applications that we have identified. An important point to make is that at a cost of $500 per hour for manned flight, a UAS unit would only be required to fly for 30 hours to pay off its cost.
We believe that Automatic Detection of obstacles is critical to developing a commercial UAV product that could be used by the market as new standards develop to allow for Unmanned Ariel Vehicles.
The service-based model for UAS could cut the hourly rate of traditional methods by 50 percent and still retain significant margins. If we assume the same business is contracted out to UAS operators at 50 percent of the hourly cost (i.e., law enforcement cost per hour is $250 instead of $500), the annual revenue potential for UAV increases significantly. The sales of both products and services will greatly enhance the market size and potential. In summary, the military market for UAS is large and well established. Significant market potential for UAS exist in the civil and commercial markets in the future.
The NSF Center for Unmanned Aircraft Systems will play an important role in facilitating technology transfer from military to civil and commercial markets. A major barrier to entry into non-military UAS markets is satisfaction of Federal Aviation Administration regulations. New algorithms, architectures, and operational procedures are required that meet both the technical needs of the UAS applications and the regulatory requirements of the FAA. Given the large size of the military markets, industry is not fully motivated to address these issues internally. The NSF Center for Unmanned Aircraft Systems provides a means of investigating these issues that are common to the UAS industry.
The Product
See following Power Point Presentation for information about the product:
http://techtransfer.byu.edu/AvailableTechnologies/files/Embedded Vision Sensor.ppt
Patent Information:
A provisional patent covering these combinations has been applied for
Inventor:
Dr. Dah-Jye Lee, Professor, Electrical and Computer Engineering
Zhaoyi Wei, Graduate Student, Brigham Young University
Licensing Information:
Contact: Dee Anderson, Associate Director, BYU Technology Transfer Office, 801 422-3676