AUTOPILOT: Automating experiments with lots of Raspberry Pis

Abstract

Neuroscience needs behavior, and behavioral experiments require the coordination of large numbers of heterogeneous hardware components and data streams. Currently available tools strongly limit the complexity and reproducibility of experiments. Here we introduce Autopilot, a complete, open-source Python framework for behavioral neuroscience that distributes experiments over networked swarms of Raspberry Pis. Autopilot enables qualitatively greater experimental flexibility by allowing arbitrary numbers of hardware components to be combined in arbitrary experimental designs. Research is made reproducible by documenting all data and task design parameters in a human-readable and publishable format at the time of collection. Autopilot provides an order-of-magnitude performance improvement over existing tools while also being an order of magnitude less costly to implement. Autopilot's flexible, scalable architecture allows neuroscientists to design the next generation of experiments to investigate the behaving brain.

Real-time, low-latency closed-loop feedback using markerless posture tracking

“…In addition to the DeepLabCut-live package that serves as a SDK for experimenters to write custom real-time pose estimation applications, we provide three methods for conducting experiments that use DeepLabCut to provide closed-loop feedback that do not require users to write any additional code: a standalone user interface called the DLC-Live! GUI (DLG), and by integrating DeepLabCut into popular existing experimental control softwares Autopilot ( Saunders and Wehr, 2019 ) and Bonsai ( Lopes et al, 2015 ).…”

“…Autopilot is a Python framework designed to overcome problems of simultaneously collecting multiple streams of data by distributing different data streams over a swarm of networked computers ( Saunders and Wehr, 2019 ). Its distributed design could be highly advantageous for naturalistic experiments that require large numbers of cameras and GPUs operating in parallel.…”

“…We demonstrate this ability via integration of DeepLabCut-Live! into the new AutoPilot framework (Saunders and Wehr, 2019).…”

“…Furthermore, it is designed to enable easy integration of real-time pose estimation using DeepLabCut into custom software applications. We demonstrate this ability via integration of into the new AutoPilot framework ( Saunders and Wehr, 2019 ).…”

Real-time, low-latency closed-loop feedback using markerless posture tracking

“…We demonstrate this ability via integration of DeepLabCut-Live! into the new AutoPilot framework (16).…”