News
From DGC
Team Caltech submits Track A proposal
(Pasadena, June 2006) Team Caltech brings together faculty and students from the California Institute of Technology, researchers and engineers from its Jet Propulsion Laboratory (JPL), and Northrop Grumman Corporation (NGC) to develop, implement, and deliver advanced technologies that will enable new capabilities in autonomous vehicles. Our technical approach is based on the following three elements, which will make up the core of our deliverables:
-
New technologies for mission and contingency management - We believe that the dominant hurdle in the 2007 Urban Challenge will be the ability to reason about complex, uncertain, spatio-temporal environments and to make decisions that enable autonomous missions to be accomplished safely and efficiently, with ample contingency planning. We will build on expertise in high confidence decision-making and autonomous mission management at JPL to develop algorithms capable of controlling the vehicle's sensing, estimation, mapping, planning and control systems in complex and uncertain conditions, while also ensuring safe operations.
-
Real-time, optimization-based navigation - Through previous DARPA and the Air Force funding, we have developed an optimization-based approach to guidance, navigation and control (GNC) that allows our vehicle to plan and execute locally optimal paths using a sensor-driven description of its environment. We will extend this approach to handle such issues as moving vehicles, traffic laws and defensive driving. Our approach will provide robust GNC modules that can be exploited by the mission and contingency management modules in a real-time, sensor-intensive system.
-
Distributed sensor fusion, mapping, and situational awareness - We will build on Caltech and JPL experience in sensory-based navigation -- including feature classification and tracking, moving obstacle detection and tracking, visual odometry, and sensory-based mapping and localization -- and extend this prior work to the highly dynamic urban environment. We will develop a multi-layer decomposition of our sensed environment so that different levels of navigation and contingency management algorithms can operate in parallel while providing highly robust and safe operation. These modules will be compatible with a highly distributed computational architecture.
We will leverage our existing Grand Challenge vehicle, Alice, and its sensing, control, and computational infrastructure so that we can focus on those technologies which are most needed for urban autonomous driving.
To develop a vehicle that is competitive for the 2007 Challenge, we will use management techniques and software tools that were developed over the past 3 years by Team Caltech for the previous grand challenge competitions. Funds from this proposal will support a post-doctoral student, 4 graduate students, and JPL engineers. We additionally anticipate including more than 30 undergraduate students and 4--8 additional graduate students in our project, as well as increased participation by JPL, NGC, and other industry partners. JPL will provide expertise in autonomy, robotics, machine vision, simulation, software engineering and testing, leveraging years of investment by NASA, DARPA and the Army in key technologies. Northrop Grumman's participation will leverage technologies and experience gained in developing safe and reliable autonomous space and flight vehicles, both manned and unmanned.
