Project Name: Drone-Mounted Species Recognition System
Aim: Trialling drone mounted GPS recognitions system to locate and visually record feral animal species that contribute negatively to total grazing pressure
Project Duration: October 2022 - July 2023
Funding body: This project is funded by the Australian Government’s Agricultural Innovation Hubs Program, with support from the Grower Group Alliance and the South-West WA Drought Resilience Adoption and Innovation Hub.
Aim: Trialling drone mounted GPS recognitions system to locate and visually record feral animal species that contribute negatively to total grazing pressure
Project Duration: October 2022 - July 2023
Funding body: This project is funded by the Australian Government’s Agricultural Innovation Hubs Program, with support from the Grower Group Alliance and the South-West WA Drought Resilience Adoption and Innovation Hub.
Executive Summary
This project aimed to develop a drone mounted species recognition system to enable us to locate and visually record feral animal species that contribute negatively to total grazing pressure.
Total grazing pressure is the ratio of the demand for pasture and the supply of pasture available. Demand can come from livestock as well as native or feral animals.
GPS positioning of feral animals will allow economic removal in the short term by pastoralists or biosecurity groups while long-term monitoring will provide a data base of feral animal movements and their interactions with other livestock and their environment.
This drone technology will allow pastoralists and biosecurity groups early warning of potential feral animal incursions. Currently pastoralists and biosecurity groups rely mostly on sightings by individual persons to detect feral animals and declared pests.
This usually occurs on foot or from a ground vehicle, for example during mill runs to check on water supplies, or other opportunistic sightings. The limited use of fixed wing aircraft or helicopters usually only occurs at mustering time, and at such times the primary focus is always on livestock that need to be mustered, and rarely on feral animals.
There is significant advantage in utilising drone technology to improve drought resilience in the Southern Rangelands. It will be particularly important in times of low feed availability such as low rainfall years or drought since it has the capacity to greatly assist in the management of non-domestic grazers and thus to enable better rangeland management.
The use of drones, which can cover more distance and inaccessible terrain, will allow more comprehensive monitoring of feral animals and in less time than traditional ground coverage requires.
Total grazing pressure is the ratio of the demand for pasture and the supply of pasture available. Demand can come from livestock as well as native or feral animals.
GPS positioning of feral animals will allow economic removal in the short term by pastoralists or biosecurity groups while long-term monitoring will provide a data base of feral animal movements and their interactions with other livestock and their environment.
This drone technology will allow pastoralists and biosecurity groups early warning of potential feral animal incursions. Currently pastoralists and biosecurity groups rely mostly on sightings by individual persons to detect feral animals and declared pests.
This usually occurs on foot or from a ground vehicle, for example during mill runs to check on water supplies, or other opportunistic sightings. The limited use of fixed wing aircraft or helicopters usually only occurs at mustering time, and at such times the primary focus is always on livestock that need to be mustered, and rarely on feral animals.
There is significant advantage in utilising drone technology to improve drought resilience in the Southern Rangelands. It will be particularly important in times of low feed availability such as low rainfall years or drought since it has the capacity to greatly assist in the management of non-domestic grazers and thus to enable better rangeland management.
The use of drones, which can cover more distance and inaccessible terrain, will allow more comprehensive monitoring of feral animals and in less time than traditional ground coverage requires.
Feral animal location and control will:
- reduce damage to fencing and water point infrastructure by large feral herbivores such as camels;
- increase livestock productivity through better management of total grazing pressures; and
- improve livestock welfare by identifying and locating feral predators such as wild dogs and feral cats.
This project will demonstrate that drones equipped with the systems we have at our disposal can provide:
- early warning of feral animal incursion into pastoral and agricultural properties; and
- production of a data base of feral animal movements and their damage to the environment and their interaction with other livestock.
Accurate data on feral animals, including GPS positions, movements, and interactions with other feral and native animals and livestock will:
- increase awareness and preparedness to potential economic and environmental issues;
- reduce disease transmission to livestock; and
- reduce potential biosecurity breaches into the food supply chain.
Data sharing of positive economic outcomes from feral animal monitoring and control will lead to faster and more widespread uptake of new innovative biosecurity programs with flow on effects into improved animal welfare and reduced land degradation issues within the whole of the Agricultural industry.
The information that can be provided by the drone technology developed through this project will be of particular benefit to biosecurity groups such as the Meekatharra Rangelands Biosecurity Association who conduct a Large Feral Herbivore cull once every three years. One of the most critical elements of planning such a cull is local knowledge of resident, or current, populations of camels, donkeys and wild horses.
Knowing where to focus the efforts of the aircraft and crew allows for maximum effectiveness of the cull program. At an average cost of $12,000 per day, if time needs to be spent scanning an area that reveals no target animals it becomes a very costly exercise for no result.
The use of a drone to scan local areas ahead of time could provide substantial savings for such major operations.
The information that can be provided by the drone technology developed through this project will be of particular benefit to biosecurity groups such as the Meekatharra Rangelands Biosecurity Association who conduct a Large Feral Herbivore cull once every three years. One of the most critical elements of planning such a cull is local knowledge of resident, or current, populations of camels, donkeys and wild horses.
Knowing where to focus the efforts of the aircraft and crew allows for maximum effectiveness of the cull program. At an average cost of $12,000 per day, if time needs to be spent scanning an area that reveals no target animals it becomes a very costly exercise for no result.
The use of a drone to scan local areas ahead of time could provide substantial savings for such major operations.
Methodology:
The methodology of this project included Planning, Design, Manufacture, Implementation, Testing and Review, with modifications and alterations applied as required in each stage, and at every iteration.
Each phase consisted of a functional specification document outlining the key objectives and design obligations. During the ensuing design phase ECR (Engineering Change Requests) were documented to show the evolution of this critical phase. (Note: these documents remain confidential to safeguard all rights belonging to the designers of each phase.) Key areas included installation guides, wiring diagrams and software features to activate and configure for communication and control.
During the testing phase data captured during site trials was utilised back in the office to streamline the feature recognition protocol. During testing ECR’s were utilised to document software and hardware improvements that were reviewed by the engineering team.
Between all major outputs updates were implemented as required from the Urgency list of the ECR’s. This will enable the product to evolve into a commercially viable product in the future.
All planning, design and manufacture was carried out at Autonomous Technology’s facilities, while testing was conducted on three properties in the Murchison region.
Each phase consisted of a functional specification document outlining the key objectives and design obligations. During the ensuing design phase ECR (Engineering Change Requests) were documented to show the evolution of this critical phase. (Note: these documents remain confidential to safeguard all rights belonging to the designers of each phase.) Key areas included installation guides, wiring diagrams and software features to activate and configure for communication and control.
During the testing phase data captured during site trials was utilised back in the office to streamline the feature recognition protocol. During testing ECR’s were utilised to document software and hardware improvements that were reviewed by the engineering team.
Between all major outputs updates were implemented as required from the Urgency list of the ECR’s. This will enable the product to evolve into a commercially viable product in the future.
All planning, design and manufacture was carried out at Autonomous Technology’s facilities, while testing was conducted on three properties in the Murchison region.
Project Activities:
The project commenced with test flights of the drone for payload and endurance capabilities. Following this the camera was mounted to the system and local testing was carried out. This demonstrated that animals can be clearly identified from a height of 40 metres and established as a heat source from over 70 metres.
Unfortunately, the project partner with whom we had planned to further develop and test the Species Recognition Software was ultimately not able to provide this software for the project. This resulted in the project continuing with a focus on visual data, thermal imaging and GPS location tagging.
Cattle were used to assess the best heat mapping colour palettes. We had to consider background “noise” and radiation in this assessment, with Black/Red ultimately proving the most effective.
Unfortunately, the project partner with whom we had planned to further develop and test the Species Recognition Software was ultimately not able to provide this software for the project. This resulted in the project continuing with a focus on visual data, thermal imaging and GPS location tagging.
Cattle were used to assess the best heat mapping colour palettes. We had to consider background “noise” and radiation in this assessment, with Black/Red ultimately proving the most effective.
Clear thermal imagery of cattle using the Black/Red palette.
The third field trip was planned to locate large feral herbivores and macropods, but due to recent rains in the area these herbivores had dispersed beyond the project area. Similarly, attempts to detect wild dogs were futile since wild dogs follow the movements of their prey species – these being predominantly herbivores and macropods.
Instead, we focused on small animals such as birds and rabbits which proved to be clearly identifiable. We reasoned that if we can identify and profile such small creatures, then larger wild dogs, feral cats and large feral herbivores will definitely also be possible.
Instead, we focused on small animals such as birds and rabbits which proved to be clearly identifiable. We reasoned that if we can identify and profile such small creatures, then larger wild dogs, feral cats and large feral herbivores will definitely also be possible.
Birds observed in trees at night (heat map image)
Ambient heat can be seen on the ground area where kangaroos have been lying down (or even standing) for long periods of time.
Conclusions:
This project has given us greatly increased knowledge and learnings of integrating autonomous systems into animal identification and control.
The project successfully conducted the planned trials of drone-mounted recognition for different classes of animals – cattle, kangaroos, birds and rabbits – on Southern Rangelands pastoral leases. The results have been very successful with imaging achieved to a very good resolution, particularly for the larger species.
Autonomous drones with species recognition, location and monitoring systems will increase the efficiency of a national biosecurity system by offering a low-cost alternative to current expensive helicopter and fixed wing feral animal location and destruction systems.
This will enable diseased animals and excessive populations to be controlled before they become an economic or environmental issue for pastoral land managers.
Data sharing of positive economic outcomes from feral animal monitoring and control will lead to faster and more widespread uptake of new innovative biosecurity programs with flow-on effects into improved animal welfare and reduced land degradation issues within the whole of the Agricultural industry.
The project successfully conducted the planned trials of drone-mounted recognition for different classes of animals – cattle, kangaroos, birds and rabbits – on Southern Rangelands pastoral leases. The results have been very successful with imaging achieved to a very good resolution, particularly for the larger species.
Autonomous drones with species recognition, location and monitoring systems will increase the efficiency of a national biosecurity system by offering a low-cost alternative to current expensive helicopter and fixed wing feral animal location and destruction systems.
This will enable diseased animals and excessive populations to be controlled before they become an economic or environmental issue for pastoral land managers.
Data sharing of positive economic outcomes from feral animal monitoring and control will lead to faster and more widespread uptake of new innovative biosecurity programs with flow-on effects into improved animal welfare and reduced land degradation issues within the whole of the Agricultural industry.
