MINE AUTOMATION
SafeAI argues that camera-based perception allows the additional classification of objects and vehicles to respond more appropriately and independently in various circumstances
Live perception data captured from the cameras is fed directly into the vehicle ’ s onboard computer so that situational assessment and navigation decisions can be made in real time and with other sensor data inputs .
SafeAI argues that there are significant advantages to camera-based perception . Autonomy 1.0 only allows for the detection of obstacles . Camera-based perception allows the additional classification of objects and vehicles to respond more appropriately and independently in various circumstances . “ For example , detected rocks in the haul road may require the truck to either stop or change its path to avoid it , but if wildlife is detected , the truck could honk and flash its lights as it approaches to motivate the wildlife to move , then only stop if required .”
Dropped material from loaded haul trucks can pose obstacles that would cause autonomous vehicles to stop . Using camera-based perception sensors , vehicles can safely navigate around dropped material and continue assigned tasks without stopping or delaying productivity .
“ Camera-based perception can be used to pack the material dumped by the vehicles more tightly and efficiently . By assessing the volumes of the existing material and the material type , the system can determine where to position the dump spot point to ensure the vehicle dumps as close to the existing material as feasible .”
Incidents in autonomous mining have occurred when map construction hasn ’ t matched current ground conditions . “ Camera-based perception can provide real-time semantic information , improving safety in areas where reliance on constructed map information can be insufficient . For example , dumping at a tip-head can depend on the map being highly accurate to allow the truck to reverse correctly to the edge . However , if the map defines a boundary that is over the edge , a truck without advanced perception could reverse over the edge , believing it has a more drivable area than it does . Perception systems would observe the ground truth position of the edge and only allow the truck to reverse right up to the edge , but not over it , despite what the configured map may define .”
Zoned-off areas for geology or maintenance typically require continual area surveyance and updating of autonomous maps . Camera-based perception can remove the need for continuous map updates to control behaviour ; vehicles can observe and apply the rules as needed .
Halder also argues that robust camera-based perception algorithms perform much better in dusty conditions , reducing false positives , which LiDAR and radar often perceive as ghost objects . “ Camera-based perception tends to provide much higher-quality , more reliable , and accurate information for the local planner , thereby increasing net availability and reducing disengagement .”
Cameras are less expensive than LiDAR and radar systems . This cost-effectiveness makes it feasible to deploy multiple cameras , providing a 360-degree view of vehicles and machinery without a significant increase in financial investment . Cameras can be easily integrated with existing hardware and software systems , allowing for a more scalable approach to upgrading and enhancing autonomous systems .
Halder concludes : “ While cameras lead in visual detail , LiDAR and radar fill critical roles by providing accurate depth information and functioning reliably under conditions where visual data might be compromised , such as in dust , fog , or extreme darkness . By integrating camera visuals with depth data from lidar and positional accuracy from radar , Autonomy 2.0 systems achieve a robust , multi-dimensional perception capability . This sensor fusion ensures greater reliability and
safety by compensating for each technology ' s limitations .”
ABD Solutions launches mobile autonomous demonstrators
The mining industry is no stranger to the advantages of autonomous technology . From enhancing safety and boosting efficiency to cutting operational costs , the benefits are undeniable . However , the leap from recognising these benefits to implementing autonomy on-site is a significant one . Mine operators are understandably cautious , wanting to ensure that any new technology will work seamlessly within the specific conditions of their site . To smooth the transition , ABD Solutions has launched a fleet of mobile autonomous demonstrator vehicles , designed to bring its Indigo Drive autonomy solution directly to mines . The fleet will be rolled-out from the beginning of October and will initially consist of demonstrators based in the UK , Japan and Australia .
The demonstrators use Polaris ATVs as the base vehicle , allowing them to be compact enough to be loaded on to a standard trailer and towed on public roads . The vehicles can be readily deployed to any site , allowing mine operators to experience autonomy in action without disrupting ongoing operations . Each vehicle has been retrofitted with Indigo Drive Auto – ABD Solutions ’ full end-to-end autonomy solution , including its integrated market-leading radar , lidar and camera-based perception system . The Polaris ’ vehicle controls are actuated using robotics that also allow them to be driven manually .
ABD Solutions told IM that these mobile demonstrators are more than just a showcase - they offer mine operators the opportunity to interact directly with the technology . They are controlled by a compact portable version of our supervisory system , which enables operators to monitor and control the autonomous
One of ABD Solutions autonomous demonstrator vehicles
34 International Mining | SEPTEMBER 2024