What is ANVEL?
The Autonomous Navigation Virtual Environment Laboratory (ANVEL) is an interactive, real-time engineering modeling and simulation (M&S) software tool built specifically to assist in the research, design, testing, and evaluation of intelligent ground vehicles. It supports the development of autonomy, semi-autonomy, and Advanced Driver Assistance Systems (ADAS) functionalities. ANVEL allows users to quickly and easily construct virtual vehicle models; place virtual sensors onboard those models; connect sensing, perception, control algorithms, and code and plugins directly to the system; and perform testing in an almost infinite variety of virtual environments with scalable fidelity. It provides scenario-focused scripting and the ability to explore performance "edge cases".
M&S for Vehicle Development
ANVEL provides virtual tools to help build a better system.
Ground vehicles are quickly evolving to become safer, more efficient, and more automated. Leveraging a combination of advanced computing, sensors, and complex algorithms, motorized ground transportation “smart” vehicles make an impact in the ground vehicle space but bringing these systems to reality poses challenges such as:
- What is the best way to develop and test my software infrastructure before the hardware platform and components are available?
- How do I test autonomy algorithms on my office desktop, to ensure they are operational before installation on a live vehicle and real-world testing?
- How can I test and refine my ADAS functions and interfaces with human subjects in a logical, controlled fashion from the safety of an office?
- How do I test my sensor code to verify the interface provides the needed data, even when the sensor is back-ordered or otherwise unavailable for use?
- How can scripting be used for exploring scenarios and benchmarking performance?
- How can I perform exhaustive testing, including edge cases, and gather performance data in an efficient, cost-effective fashion?
These challenges and questions are the heart of the intelligent vehicle development and testing trials in which ANVEL can help.
ANVEL contains user-friendly functions and tools, including:
- Real-time performance to facilitate teleoperation, shared control, and autonomous behaviors; faster than real-time to facilitate larger-scale experimentation; slower than real time to facilitate higher fidelity, more computationally intensive simulation components
- Inherently multi-fidelic – use higher fidelity simulation components in areas of deeper interest and lower fidelity components where there is less importance to yield a complete systemic simulation; choose the fidelity throughout, giving it the flexibility to address many challenges
- Included preconfigured environments, assets, and vehicle and sensor models that users can manipulate to define new ones
- Integrates directly with algorithms, code, plugins and programming languages (e.g., Python, C, C++, Matlab, etc.)
- Interface with and support for interactions with other external systems via the External Application Program Interface (API) which provides the ability to programmatically query, configure, or control anything within the ANVEL simulation and enable possibilities for testing and validation; ANVEL’s external API supports common languages, including Python, C, C++, and Matlab
- Interfaces with external hardware and software systems to form flexible hardware-in-the-loop and software-in-the-loop combinations
- Quick and easy creation of virtual vehicle models of different design and architectures, using easy-to-edit XML files to define details and graphical models from common digital modeling tools
- Simple importing, editing, and configuration of environments, including modifying terrain heights, changing terrain composition (grass, gravel, asphalt, etc.), and manual and procedural insertion/placement of objects
- Easily attach virtual sensors to a vehicle or other assets with point-and-click tools; used as part of the overall simulation and to support by-wire accurate representations of protocols
- Ability to incorporate sensing, perception, and control algorithms/code and autonomy codebases, integrate them with the virtual vehicle/sensors/systems, and perform testing and experiments
- Instrumentation of variables and systems within the simulation, capturing performance data from anything within the vehicle or environment, allowing real time plotting and exporting
- Support for virtual “actors” that interact with the system under test; virtual vehicles, humans, animals, and other assets can be spawned in real time, follow paths or scripts, respond to triggering events, and change behaviors
ANVEL’s capabilities are growing to support the expanding community and breadth of applications. Quantum Signal (developer of ANVEL) uses the tool to develop a host of technologies and systems for manned and unmanned ground vehicle development, as do hundreds of its users throughout the world.
Supporting Multiple Efforts
ANVEL spans the entire systems engineering process.
ANVEL began life supporting military efforts developing and testing unmanned ground vehicles in complex, challenging off-road environments. Since then, ANVEL has grown and evolved into a substantial, real-time, mixed-fidelity, highly-flexible tool that supports a wide variety of modeling and simulation capabilities for intelligent ground vehicles. This growth has come out of both design and necessity; as the requirements and demand for intelligent ground vehicles have increased, so too has the need for M&S capability to support this growth. ANVEL has been, and is being, used to address areas such as:
- Exploration of concept of operation (CONOPS) for vehicle systems with different configurations operating in a variety of environments; facilitating interactive exploration of CONOPS helps developers define and distill system requirements.
- Exploration, testing, and debugging of autonomy and related algorithms, including path planning, obstacle detection and avoidance, and discrete behaviors.
- Exploration, prototyping, and refinement of user interfaces; ANVEL acts as the virtual unmanned ground vehicle while different designs for operator control units and user interfaces are tested with human operators
- Testing and debugging of X-by-wire systems, sensor and actuator interfaces, autonomy controller boxes, and other subsystems to refine and tune their performance, individually or together as a system
- Development of HIL/SIL systems; developer of ANVEL (Quantum Signal), the US Army, and others use ANVEL to connect real-time virtual components (vehicles, sensors, etc.) to real world components in various combinations to test software and algorithm systems and subsystems prior to deployment on live platforms
- Testing and characterization of the performance of the systems; this includes large scale, repeated testing of algorithms or software components while varying parameters within the simulation and generating outputs that can be further analyzed and leveraged.
- Exploration of multi-vehicle/convoy systems; simulation of leader-follower behaviors and “platooning” are possible in ANVEL; Quantum Signal is working with the US Army to leverage this capability in multiple programs