
JAUS - Software Development ToolsTransport Layer Support and Low Level Message Handling: jNM and jLib support message transport via shared memory, Ethernet, or serial links, with the exact transport layer to be used selectable at runtime. jNM and jLib effectively present a single low latency interface for any component to communicate with any other component in the JAUS system while maintaining a single send and receive messaging API. Ethernet and serial communication are supported for inter-node and inter-subsystem communication between Node Manager components – including multiple instances of jNM. Finally, jNM can simultaneously support all three types of transport interfaces and route messages appropriately between each of them. Message Routing: jNM supports both component-to-component and broadcast routing of messages within the full scope of the JAUS message ID. When jNM receives a component-to-component message, it compares the destination ID with the destination ID of all known data links. If a complete match is found (subsystem, node, component, and instance all match), the message is forwarded directly to the destination component. If a complete match is not found, jNM checks to see if it has any data links to other jNM’s whose node and/or subsystem ID matches the destination ID of the message. If so, it forwards the message to that jNM for eventually delivery to the appropriate destination component. This message forwarding will happen continually, until the message reaches its final destination component. If jNM receives a message with a broadcast ID in place of the subsystem, component, node and/or instance ID, it uses the same rules described above to choose the appropriate set of destinations to forward the message to. In addition, each jNM has Broadcast Loop Canceling algorithms that prevent a broadcast message from being endlessly circulated throughout the JAUS network. Dynamic Topology and Capability Discovery: One of the unique features of jNM and jLib is their ability to automatically find and connect to components within a node as well on different nodes. Practically, this means that when a component starts it doesn’t need to know specifics of how to connect to the Node Manager on its node – the actual connection method is handled within jLib. From a system level, this means that the entire JAUS system doesn’t need to be specified a-priori, which is a significant issue for large and/or dynamic JAUS systems. After initializing jNM with its subsystem and node id, and each other component with its component and node id, no further network configuration by the user needs to take place – full configuration and routing happen automatically – even as components come and go from the system because of radio dropouts, software glitches, or intentionally shutdowns. Similarly, jLib supports the automatic reporting of the capabilities of each component with the JAUS system through the event notification protocol. When a new JAUS component enters the system, all registered components will be notified about the new component’s network address and the capabilities that it provides. In this way, JAUS components can optimally connect and disconnect from appropriate system resources as they become available. Component Control and Authority Limitations: jLib as well as all appropriate JAUS capability components that Applied Perception sells support the JAUS component control and authority constructs. These constructs limit which components can control (send messages that are acted upon) other components based on each component's user-defined authority level, as well as prevent multiple components from simultaneously controlling another component at the same time. These features are particularly important for components that control basic platform mobility (Primitive Driver) or mission/safety critical devices such as weapons systems. Service Connections: Periodic messages are a fundamental building block of nearly all unmanned systems. From the beginning JAUS has supported periodic messages through the use of the JAUS Service Connection construct. jLib as well as all of Applied Perception’s JAUS capability components support the creation, maintenance, and teardown of multiple service connections per JAUS component. By using this construct system developers can dramatically cut down the number of JAUS query messages that must be sent in order to receive period information, such as from Global, Local, or Visual Sensor components. JAUS Payload Interface: Because the number of possible payloads that could be added to an unmanned vehicle is almost limitless, the JAUS community developed a generic payload specification that would enable any JAUS-compliant graphical operator control unit to control any on-board payload. jLib as well as our JAUS-compliant operator control unit support this payload interface, which allows developers to quickly and easily customize there unmanned system without needing to create additional operator control interfaces. Rather, by using the JAUS Payload interface, nearly any payload can be seamlessly integrated. Network Monitoring: jNM can be configured to optionally track every message based on source, destination, and message ID. This data is used for generating statistics about message traffic between components and nodes. |
PRESS RELEASES:
![]()
SkunkWorks: The TARDEC Skunk Work project is centered on the TAGS-CX vehicle platform. This platform is a high-performance, high-speed, high-mobility JAUS-compliant semi-autonomous unmanned ground vehicle (UGV) with modular payload capabilities. It is the result of a collaborative design process involving personnel from Applied Perception Inc., TARDEC, and Dumur Industries.
![]()
RCCEE: Robotic Combat Casualty Extraction and Evacuation Overview: Robotics technologies that can seek, detect, extract and transport a wounded soldier from the point of injury to a field hospital. Numerous sensing, navigation, and other technologies have been researched and developed during this program.
![]()
LAGR: The DARPA LAGR program aims to develop completely autonomous off-road ground vehicle navigation technologies using only computer vision and machine learning techniques.
|
|