Low power neural network-based control and actuation solutions for insect-scale robots
Date
2024-06-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Kujutage ette maailma, kus tillukesed, putukasuurused, robotid suudavad autonoomselt täita ülesandeid nagu avastamata alade uurimine, otsingu- ja päästemissioonidel abistamine või keskkonna seire. Need väikesed robotid on mitte ainult kulutõhusad tänu oma minimaalsetele materjalinõuetele, vaid neil on ka potentsiaal tegutseda suurtes kogustes laiaulatusliku katvuse saavutamiseks.
Väljakutse seisneb selles, et varustada neid putukasuuruseid roboteid võimega end ise toita ja oma liikumist kontrollida ilma välise abita. Traditsioonilised komponendid on liiga energiamahukad ja lisavad kaalu, mis on selliste väikeste robotite jaoks ebapraktiline. Seetõttu on putukasuurused robotid sageli piiratud väliste toiteallikate ja juhtimismehhanismidega, mis piiravad nende tööulatust.
Käesolev dissertatsioon tutvustab uuenduslikke lahendusi, et võimaldada nendel robotitel saavutada autonoomsust. Uuritakse madala võimsusega pehmete täiturite kasutamist, mis on inspireeritud looduslikest liikumistest, võimaldades robotitel roomates energiat säästa. Lisaks pakutakse välja uusi energiasäästlikuid elektroonikaskeeme visuaalse teabe töötlemiseks, võimaldades robotitel oma keskkonnas iseseisvalt navigeerida.
Uurimistöö märkimisväärne saavutus on skeemi väljatöötamine, mis suudab erakordselt madala energiatarbimisega kujutist tuvastada. See edasiminek on oluline samm autonoomsete putukamõõtmeliste robotite, mis suudavad erinevates keskkondades mitmesuguseid ülesandeid täita, realiseerimise suunas.
Uurimistöö mõjud ulatuvad kaugemale praegusest tehnika tasemest, seades aluse tulevastele edusammudele miniatuurse robotitehnika valdkonnas. See avab võimalused nende robotite parvede kasutuselevõtuks stsenaariumides, kus nende väike suurus ja autonoomia saab suurepäraselt ära kasutada, kuulutades välja uue ajastu robotitehnika alal.
Envision a world where tiny robots, as small as insects, could autonomously perform tasks like exploring uncharted territories, assisting in search and rescue missions, or conducting environmental surveillance. These minuscule robots are not only cost-effective due to their minimal material requirements but also have the potential to operate in large numbers for extensive area coverage. The challenge lies in equipping these insect-scale robots with the ability to power themselves and control their movements without external assistance. Traditional components are too power- intensive and add extra weight, which is impractical for such small robots. Consequently, insect- scale robots are often restricted by external power sources and control mechanisms, limiting their operational range. This thesis introduces innovative solutions to empower these robots with self-sufficiency. It explores the use of low-power soft actuators inspired by natural movements, allowing the robots to conserve energy by crawling. Additionally, it proposes new energy-efficient circuits for processing visual information, enabling the robots to navigate their environment independently. A notable achievement of this research is the development of a circuit capable of image recognition with exceptionally low power consumption. This advancement is a significant step towards the realization of autonomous insect-scale robots capable of undertaking diverse tasks in various environments. The implications of this research extend beyond the current scope, setting the stage for future advancements in the field of miniature robotics. It opens up possibilities for deploying swarms of these robots in scenarios where their small size and autonomy can be leveraged to great effect, heralding a new era in robotics.
Envision a world where tiny robots, as small as insects, could autonomously perform tasks like exploring uncharted territories, assisting in search and rescue missions, or conducting environmental surveillance. These minuscule robots are not only cost-effective due to their minimal material requirements but also have the potential to operate in large numbers for extensive area coverage. The challenge lies in equipping these insect-scale robots with the ability to power themselves and control their movements without external assistance. Traditional components are too power- intensive and add extra weight, which is impractical for such small robots. Consequently, insect- scale robots are often restricted by external power sources and control mechanisms, limiting their operational range. This thesis introduces innovative solutions to empower these robots with self-sufficiency. It explores the use of low-power soft actuators inspired by natural movements, allowing the robots to conserve energy by crawling. Additionally, it proposes new energy-efficient circuits for processing visual information, enabling the robots to navigate their environment independently. A notable achievement of this research is the development of a circuit capable of image recognition with exceptionally low power consumption. This advancement is a significant step towards the realization of autonomous insect-scale robots capable of undertaking diverse tasks in various environments. The implications of this research extend beyond the current scope, setting the stage for future advancements in the field of miniature robotics. It opens up possibilities for deploying swarms of these robots in scenarios where their small size and autonomy can be leveraged to great effect, heralding a new era in robotics.
Description
Väitekirja elektrooniline versioon ei sisalda publikatsioone