Tiny-PULP-Dronets: Squeezing Neural Networks for Faster and Lighter Inference on Multi-Tasking Autonomous Nano-Drones

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lamberti, Lorenzo, Niculescu, Vlad, Barcis, Michał, Bellone, Lorenzo, Natalizio, Enrico, Benini, Luca, Palossi, Daniele
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909239400726528
author Lamberti, Lorenzo
Niculescu, Vlad
Barcis, Michał
Bellone, Lorenzo
Natalizio, Enrico
Benini, Luca
Palossi, Daniele
author_facet Lamberti, Lorenzo
Niculescu, Vlad
Barcis, Michał
Bellone, Lorenzo
Natalizio, Enrico
Benini, Luca
Palossi, Daniele
contents Pocket-sized autonomous nano-drones can revolutionize many robotic use cases, such as visual inspection in narrow, constrained spaces, and ensure safer human-robot interaction due to their tiny form factor and weight -- i.e., tens of grams. This compelling vision is challenged by the high level of intelligence needed aboard, which clashes against the limited computational and storage resources available on PULP (parallel-ultra-low-power) MCU class navigation and mission controllers that can be hosted aboard. This work moves from PULP-Dronet, a State-of-the-Art convolutional neural network for autonomous navigation on nano-drones. We introduce Tiny-PULP-Dronet: a novel methodology to squeeze by more than one order of magnitude model size (50x fewer parameters), and number of operations (27x less multiply-and-accumulate) required to run inference with similar flight performance as PULP-Dronet. This massive reduction paves the way towards affordable multi-tasking on nano-drones, a fundamental requirement for achieving high-level intelligence.
format Preprint
id arxiv_https___arxiv_org_abs_2407_02405
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Tiny-PULP-Dronets: Squeezing Neural Networks for Faster and Lighter Inference on Multi-Tasking Autonomous Nano-Drones
Lamberti, Lorenzo
Niculescu, Vlad
Barcis, Michał
Bellone, Lorenzo
Natalizio, Enrico
Benini, Luca
Palossi, Daniele
Robotics
Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
Pocket-sized autonomous nano-drones can revolutionize many robotic use cases, such as visual inspection in narrow, constrained spaces, and ensure safer human-robot interaction due to their tiny form factor and weight -- i.e., tens of grams. This compelling vision is challenged by the high level of intelligence needed aboard, which clashes against the limited computational and storage resources available on PULP (parallel-ultra-low-power) MCU class navigation and mission controllers that can be hosted aboard. This work moves from PULP-Dronet, a State-of-the-Art convolutional neural network for autonomous navigation on nano-drones. We introduce Tiny-PULP-Dronet: a novel methodology to squeeze by more than one order of magnitude model size (50x fewer parameters), and number of operations (27x less multiply-and-accumulate) required to run inference with similar flight performance as PULP-Dronet. This massive reduction paves the way towards affordable multi-tasking on nano-drones, a fundamental requirement for achieving high-level intelligence.
title Tiny-PULP-Dronets: Squeezing Neural Networks for Faster and Lighter Inference on Multi-Tasking Autonomous Nano-Drones
topic Robotics
Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
url https://arxiv.org/abs/2407.02405