Achieving true autonomy, especially in GPS-denied environments (e.g. indoors, underground, etc.) is a very challenging problem.
True autonomy in the world of technology and computing originates from the human autonomous nervous systems and is based around the properties of self-configuration, self-optimization, self-healing, and self-protection.
The self-protection property implies the system’s ability to adapt to a dynamically changing environment; self-optimization allows a system to tune resources and balance workload to maximize their use; self-healing is the ability to discover, diagnose, and act to prevent disruptions; and, self-protection enables the anticipation, detection, identification, and protection against attacks.
Based on these properties it is evident that very few if any unmanned systems have yet earned claims of autonomy, however, Robotics Centre’s focus on autonomy requires at a minimum the following autonomous properties vital to independent push-button operation of unmanned systems:
Mission planning and execution
The ability for an unmanned system to not only navigate independently but plan and execute missions in real-time. To this end, an operator should only provide the unmanned system with a high-level mission (e.g. move from point A to point B). The unmanned system itself is responsible for all low-level planning in terms of navigation.
Real-time situational awareness
The ability for an unmanned system to sense and perceive the 3D environment in real-time. Should its path become obstructed, the unmanned system senses this obstruction in real-time and re-plans accordingly. The unmanned system does not need a human operator to intervene nor provide the unmanned system with new information if its path is blocked.
All intelligence onboard
All of the computation needs of the unmanned system must exist onboard. There should be no need for edge or cloud computing to process data, which leads to the next property.
Similarly to the property above, the unmanned system has no requirement for network or radio infrastructure in the execution of a mission.
The unmanned system should possess an intelligent software architecture allowing for the ability for a set of multiple unmanned systems to work together or independently to complete a large mission (too large for a single robot to complete)
Multi spectral sensor fusion
Similar to how the autonomous nervous system utilizes five senses, the unmanned system should utilize a combination of sensors (E.g. LiDAR, IMU, camera) to perceive the environment. Loss of one of these sensors, may downgrade the capability of the system but not leave it nonoperational.