We are proponents of using a Model Based Systems Engineering (MBSE) approach to the development of autonomous vehicles, and as part of this approach, we rely on the Virtual Autonomous Vehicle Simulation (VAVS) TDP which is intended to be platform that can be used for simulating an individual vehicle or a fleet of vehicles, manually or autonomously driven, in varied traffic situations and road conditions.
Once initial development is done using the VAVS-TDP it is natural to move to further development and testing with real vehicles using real hardware. However development work on real autonomous vehicles poses potential high safety risks, and therefore development can happen only in very constrained scenarios. These constraints are further exacerbated when we want to introduce new students to real software and hardware (sensors, actuators, compute platform, software). Further the environment where the scenarios can be executed become much less controlled (weather, traffic, etc) making the development happen at a much slower pace.
The Scaled Autonomous Vehicle Indoor (SAVI) Technology Demonstrator Platform is intended to address the challenges of going directly from a virtual world to full-scale vehicles. The SAVI-TDP will consist of multiple scaled and radio-controlled cars, and an indoor scaled city environment. The scaled vehicles will have advanced sensors used in autonomous vehicles, such as LIDARs, Cameras, IMUs, etc. They will have advanced compute platforms that can run software comparable to that used in real vehicles. The scaled city will consist of mock buildings, as well as connectivity that can simulate some of the connectivity that can be available in outdoor environments. The SAVI-TDP will become the basis of evaluation of several autonomous vehicle technologies through individual projects.
The SAVI-TDP will be a key learning platform for students, as students build / assemble / integrate hardware and software on specific projects as well as student competitions.
The development of the SAVI-TDP itself is expected to uncover new research challenges. In particular, how does the scaling of the cars (both in terms of the vehicle dynamics as well as sensor spatial scaling) affect the performance of the autonomous algorithms, and how can that be mitigated? Again such research questions will be addressed through individual projects.
- Multiple Scaled RC Vehicles have been built
- An indoor lab has been converted to a scaled city with mock buildings, road, lane markings, etc. The lab has been equipped with stationary communication devices, cameras, etc. to provide some of the connectivity and localization capability available in outdoor environments.
- Multiple projects are on-going leveraging this TDP
Projects Leveraging the TDP
- 1/10th Scaled Car High Speed Waypoint Following This project focuses on creating a 1/10th scale platform capable of high speed waypoint following. This will allow testing of control algorithms in a low risk environment.