ReachingBot

autoscientific ReachingBot®

ReachingBots automate the single pellet reach and grasp task for both mice and rats, the gold-standard test of dexterity and grasping. ReachingBots are benchtop devices.

Highly parallel training and assessment

ReachingBots enable researchers to train an unlimited number of animals in parallel: the software handles the hard part for you. Plug your devices into a hub, connect them to your PC, then click go.

High-speed video recording

ReachingBots leverage on-board cutting-edge machine vision algorithms that detect for the presence of a dispensed pellet in real-time. The camera captures frames at 200 Hz but only saves video footage when something interesting happens. This means each video is event-synchonised, and enables users to create rich datasets to explore motor function in depth.

Designed and manufactured in the UK

ReachingBots were designed from the ground up in London, UK. We use the highest quality materials, from Nylon PA12, to precision machined anodised aluminium, ensuring our devices last.

autoscientific ReachingBot® Device Manager

TheReachingBot Device Manager (RBDM) is the main software interface used to interact with ReachingBots. The RBDM handles experimental configuration, automatic trial classification and data inspection. Browse the documentation.

Plug-and-play
The ReachingBot Device manager make autoscientific ReachingBots® plug-and-play. You won't need to program your devices or need know how to code. Simply plug your devices in, pair them with your PC, and you're ready to go.
AI-powered
Our software comes built-in with machine learning algorithms that do not require expensive hardware to use. Simply select the videos you'd like to be analysed and your normal laptop/PC will handle it.
Optimised for the whole lab
The ReachingBot device manager is highly configurable. Multiple users can share ReachingBots easily. Simply create a configuration that suits your experiments, then load it each time you're ready to use the devices.

1 3

What does ReachingBot do?

ReachingBot automates the single pellet reach-and-grasp (SPRG) task. It dispenses pellets, records event-triggered 200 Hz video, and scores every trial as a hit or miss using machine learning.

How many animals can I train at once?

As many as you have devices. In the published validation, 10 ReachingBots on about 2 m of bench trained 30 mice in parallel, recording around 2,000 trials a day.

How much time does it save compared with manual training?

In the published study, daily training took under 1.5 hours including set-up, compared with an estimated 10+ attended hours by hand. Mice also performed about 80 trials per 20-minute session, versus roughly 30 pellets in a typical manual session.

How accurate is the automatic trial scoring?

The built-in classifier scored hits and misses with 94% accuracy, running on an ordinary PC CPU with no GPU required.

Is it sensitive to motor deficits after injury?

Yes. After a photothrombotic motor cortex stroke, mean peak success rate fell from 52% to 25% and the deficit persisted throughout the 4-week assessment.

Has ReachingBot been published?

Yes: Kakanos SG, Gadiagellan D, Kim E, Cash D, Moon LDF (2023). ReachingBot: an automated and scalable benchtop device for highly parallel Single Pellet Reach-and-Grasp training and assessment in mice. Journal of Neuroscience Methods, 109908. doi:10.1016/j.jneumeth.2023.109908 (preprint on bioRxiv).

The validation study was carried out in mice. ReachingBot can also be used with rats; rat validation data is not yet published.

How do I get a quote?

Use the contact form below or email info@autoscientific.co.uk. Devices can be customised to your lab's needs.

Item added to your cart