Inside Unmanned Systems provides actionable business intelligence to decision-makers and influencers operating within the global UAS community. Features include analysis of key technologies, policy/regulatory developments and new product design.
Issue link: http://insideunmanned.epubxp.com/i/763107
42 unmanned systems inside December 2016/January 2017 V2X or connected vehicles you will need even better accuracy. This will require additional sensors," Basnayake said. "When we get to the beyond scenario, you will need even better ac- curacy. Sensor integration will be the way to do that. …Industry needs to see we are going from accuracy of several meters to applications that will require centimeter level accuracy." V2X will be an important aspect, he said, because traditional sensors have limitations and could be blocked by other objects, rain, lighting and fog. That makes it difficult to predict traffic behaviors. Connected vehicles make information sharing possible, which is a key piece to developing driverless cars that can safely travel on our roadways. Quantifying Navigation Safety As of now, there isn't enough information from companies testing driverless cars, such as Google and Tesla, to disprove or prove the safety case, Joerger said. The industry can, however, leverage what's been used in air navi- gation for decades. "When we talk about aviation, how do we evaluate safety? If we have an aircraft during an approach, this aircraft will try to predict its position close to the touchdown point and it wants its positon estimate to be within a pre- WEBINAR RECAP » DRIVERLESS CARS BY THE NUMBERS » PARTICIPANTS' VIEWS After 2025: 47 % 2020–2025: 46 % Before 2020: 7 % When do you think fully autonomous cars will be massed produced? Inertial: 27 % Map Matching: 18 % GNSS: 64 % LiDAR/Radar: 61 % Cameras: 34 % In your opinion what is the most important technology in an autonomous car? (Select your top two) Cost: 33 % Certification: 43 % Connectivity/cyber security: 50 % Sensor technology: 28 % Confidence that users will adopt: 39 % In your opinion what is the biggest challenge with autonomous cars? (Select your top two) defined requirement box. The probability of being outside that predefined box is called the integrity risk or the safety risk. If we want to translate that into the car problem then we can draw a box that is pretty much limited by the width of a lane. So the first thing that is strik- ing when looking at this is comparing the size of the boxes. It took decades of research and considerable resources in the aviation indus- try to reduce the size of the box," Joerger said. "This is an order of magnitude larger than what we need for a car. It's a task of epic pro- portions. Clearly adjustments need to be made if we want to convert these requirements using aviation to the autonomous car problem." What are some of these adjustments? Be- cause GPS is not reliably available on cars, Joerger said a multisensory system is neces- sary for navigation. And while aircraft are only monitored for risks at touchdown, driverless cars will need to be continually monitored in a dynamic environment for risk prediction. To make a full safety case in the aviation world, four criteria must be considered: accu- racy, integrity, or the measure of trust in safety information, continuity risk and availability. "We want to apply these criteria to an au- tonomous car navigation system," Joerger said, stressing the importance of integrity. "We need