The challenge
Mobile robots providing safety and security in industry across Australia
Autonomous mobile robots can provide improved safety and productivity across many industries which are of importance to Australia., such as mining, construction, agriculture, defence and forestry. Deploying teams of robots to help deliver outcomes in these industries, often requires them to operate dependably in unstructured and changing environments, where there are no external positioning systems such as GPS, or pre-existing localisation maps.
In these environments, robots must use their on-board sensors and perception algorithms to simultaneously build detailed maps of the world whilst localising themselves within them. This technology is referred to as SLAM, or Simultaneous Localisation and Mapping. Whilst there are many approaches, and a few commercially available solutions, none of them sufficiently robust, reliable or accurate to meet the needs of industry, nor directly support the operation of the robot teams through map sharing and cooperative localisation.
Our response
Algorithms convert data from robots sensors into environmental map
Wildcat is being developed by CSIRO's Data61 Robotics and Autonomous Systems Group and addresses the challenge of industrial-grade accuracy, robustness, reliability and flexibility while supporting the operation of robot teams through map sharing and cooperative localisation.
Wildcat uses advanced non-linear optimisation algorithms to convert 3D LiDAR (Light Detection and Ranging) data from the robot's sensors into highly detailed and accurate maps of an environment in real time, whilst providing update-rate pose information for robot control. Robots can share map segments in real time to assist them to maintain awareness of changes in the environment, and to explore new areas.
CSIRO’s Data61 Wildcat creating a detailed scan of an indoor environment and detecting height of objects, to act as a map for an autonomous robot
Wildcat leverages more than ten years of SLAM R&D at CSIRO, and has been designed as a research and development platform for autonomous field robots and a commercially ready solution for industrial applications.
The technology works in conjunction with an IMU (Inertial Motion Unit) and LiDAR sensors with the option to add other sensors such as chemical, radiological, gas, GPS and Wi-Fi.
Wildcat's advantages include:
- Robust Performance: It works reliably in the toughest environments such as tunnels, confined spaces and open areas, where other SLAM systems fail. The system automatically corrects for errors in large maps.
- Extreme Accuracy: Wildcat won the most accurate object detection prize in the DARPA Subterranean Challenge, a global robotics competition for underground exploration and mapping, where CSIRO competed against the best field robotics groups in the world.
- Versatile, Modular and Reconfigurable: It has an easy-to-deploy core and plug-in system to support new functionality and experimentation and can is easily adapted to work with novel sensors and sensor configurations.
- Increased Performance: Wildcat is equipped with advanced algorithms designed for use in real-world systems.
- Easy to integrate: It is ROS compatible, with a documented SDK (Software Development Kit) and offers Window support.
- Efficient Real-Time Operation: It runs with embedded CPUs, multi-process and multi-thread options, with low bandwith map sharing.
- Game-changing Features: Multi-agent SLAM enables teams of people and robots to rapidly map and explore complex environments.
Wildcat can be built into new or existing products and services to provide accurate, robust, fast, efficient localisation and mapping to navigate both built and natural environments in 3D.
The results
A versatile, robust and accurate SLAM solution
Wildcat is designed for the needs of industry, and the potential to improve safety and productivity is significant. Our goal is to work with Australian companies to commercialise the technology within different market segments.
Wildcat can provide companies with a significant technological edge that assists with driving export opportunities and jobs growth. It also supports broader Australian industry through the development of new products and services that can be sourced and supported through the domestic supply chain.
Wildcat is designed for:
- Autonomous systems
- Perception systems for autonomous robots, drones and vehicles
- Mobile mapping and localisation
- Real-time mapping and localisation for mobile applications
- Retro-fitted systems
- Additions to pre-existing sensor payloads requiring state estimation.
The Wildcat Early Adopter Program provides early access to the Wildcat technology and is a free to join opportunity for selected participants. The program enables small to medium enterprises (SMEs) to evaluate and develop new products and services using world-class and high-performance technology.
As part of the program, selected participants receive:
- 12 months non-exclusive access and use of the Wildcat software
- The option to licence our "Catpack" integrated perception system design (seen in the above image, mounted onto a Nexxis inspection robot)
- Basic integration support and basic training
- Basic ongoing support for data processing and de-bugging.
Commercial gains from utilisation of the software during the program will attract a fee.
Emesent, a drone autonomy technology company, is Wildcat's first early adopter, and are accompanied by several other companies.
For SMEs who may be unsure if Wildcat is compatible with their products, a feasibility study can be undertaken to identify opportunities. If engineering support is required for integration only, we may be able to assist via a side project.
For further information on Wildcat please contact Fred Pauling on [email protected].