MKMT has secured funding from Western Bay of Plenty District Council and the Joint Agency Partnership (DOC, WRC and BOPRC) to purchase AI traps for a year-long trial in the Kaimai Mamaku project area.

We are yet to confirm budget split in regards to how many traps need to be deployed to each rohe.

The Kaimai Mamaku terrain is tough, volunteer and paid staff time is limited and project funding is restricted.

MKMT are constantly looking for ways to improve pest animal control networks. To this end, we have been investigating the option of introducing traps enabled with artificial intelligence (AI) that recognise, arm and humanely kill pest species, while recognising and remaining disarmed for non-target species to avoid by-kill of native species and harm to people.

AI traps should save people time, effort and improve health and safety.

The purpose of the trial is to test the capability and usefulness of AI traps for conservation groups in the Kaimai Mamaku terrain and to share that knowledge for the benefit of the conservation community.

There are several AI-enabled traps on the market with advantages and disadvantages. MKMT has investigated these and found that the trap best suited to the Kaimai Mamaku groups and terrain is the Critter Solutions Multi-Trap.

Unlike other AI traps, each trap is a standalone unit with its own communications system and does not require a trap network to operate effectively. The communications provider is SpaceX which means each unit can talk to a satellite, enabling the AI learning and refinement, species identification and trap arming, and trap maintenance and function to be managed effectively in remote environments.

• Traps are set to arrive in October

• Training will begin in November (November 3rd)

Using SpaceX/Starlink satellite enabled simcard ensures your trap has reception in most locations without the need of repeaters. Data will be automatically sent to your device in a cadence you choose, and data will also be stored on a SD card within the device.  

This is an open trap to reduce any barriers for the target animal. It is designed to not kill until an animal is correctly placed, reducing the likelihood of trap shyness and the lure is not desirable for insects, reducing the likelihood of birds interacting with the trap. 

Each team will get 2-3 traps, depending on final participant numbers. The aim of this trial isn't to replace all your traps with AI, it's to figure out how to best use new technology within existing networks - may it be in a hot spot, a hard to reach area, or somewhere that predators have become trap shy. We will help you figure out where to best put this within your own networks. 

There currently isn't any written report from a large trial, but there is qualitative feedback that the traps are highly effective. We will be inviting users to our training day, to share their experiences with you.

We will create a written report, with aggregated data from all groups, at the end of the trial for the betterment of the entire conservation industry. 

Cats are not a target species for the Multi-Trap, but there is always a chance the onboard AI could incorrectly classify an animal. Although there hasn't been a non-target trigger to date, we advise that traps in urban areas are mounted up trees to reduce the chance of pets interacting with the trap.

Yes, and it will continue learning based on all trap users. There has been no evidence that birds are attracted to the trap, and the bait has been developed to not be attractive to insects. It has been tested with taxidermy kiwi beaks to ensure the AI can correctly and instantly identify kiwi and not fire.

Standard has been set to 10 minutes.

Please let MKMT know early so we can discuss options and ways to support your team. Failing to find a solution, your trap may be redeployed - all information can be found in the MOU.