Adaptive Control of Underwater Snake Robot

Abstract

Ideen om bruk av slangeroboter til undervannsoperasjoner handler om dens fleksibilitet og fordelaktige evne til å bevege seg rundt og inne i installasjoner under vann. Slangeroboter er antatt mer egnet for komplekse operasjoner enn mer konvensjonelle undervannsfarkoster.

Hovedfokuset i denne oppgaven er implementasjon av algoritmer for både adaptiv regulering og invers kinematikk for en slangerobot-simulator i MATLAB. Grunnen for at man vil bruke adaptiv regulering for slangeroboten er at den kan ha en veldig varierende dynamikk, da den kan innta flere ulike konfigurasjoner. Det er vanskelig å modellere kreftene på en kompleks robot som har en stadig varierende konfigurasjon. Derfor er det ønskelig å ha en regulator som kan tilpasse seg den stadig varierende dynamikken. Slangeroboter skal kunne utføre ulike oppgaver under vann. Invers kinematikk er viktig for å kunne generere referansene for hvordan slangen faktisk skal bevege seg for å gjennomføre oppgavene den blir satt til å løse.

Et kort litteraturstudie har blitt skrevet om både slangerobotikk, adaptiv regulering og invers kinematikk. Teorien som er nødvendig for implementasjon av algoritmene for invers kinematikk og adaptiv regulering er presentert, etterfulgt av en metodisk gjennomgåelse av hvordan systemet er definert.

Det implementerte systemet har blitt simulert og testet for fire ulike scenarioer. Utfra disse simuleringene kan det konkluderes med at den adaptive reguleringen fungerer svært bra for de aller fleste oppgavene, men har en liten svakhet. Den implementerte invers kinematikken fungerer best for de enkleste scenarioene, men har problemer med å løse flere oppgaver samtidig.

The idea of using underwater snake robots for subsea operations comes from their ability to perform a wide range of missions, due to the slender and flexible body. Because of these advantages, they are expected to perform more complicated tasks than the more conventional remotely operated vehicles.

The main focus of this thesis is the implementation of an adaptive control algorithm together with an inverse kinematics algorithm to control an underwater snake robot simulator in MATLAB. The need for an adaptive controller is identified as the snake robot might take a range of configurations and interfere with objects subsea. This means that the drag forces, among others, can vary a lot and change rapidly. Such changes are difficult to model and the idea is to have a controller which adapts to these changes. The underwater snake robot should also perform several tasks. By defining the relationship between the snake robot’s controllable states and the actual task, the inverse kinematics should be able to calculate the desired motion for the snake robot to achieve the task.

Relevant background on these topics have been presented through a literature review. An adaptive control law has been implemented for the snake robot simulator in this thesis, along with an inverse kinematics algorithm. The theory behind these algorithms is presented, and the process of implementing them is described. To implement these algorithms correctly, theory on kinematics for the snake robot have also been presented.

At last, the system has been simulated for four different cases. Each case defined one or several tasks that should be achieved. The results showed that the adaptive controller worked very well for most cases but had a small issue when the weight of the snake robot was changed to simulate changed dynamics. The inverse kinematics also performed well for the simplest cases but could not always find a feasible solution to achieve several tasks. The conclusion is that the implemented adaptive controller and inverse kinematics could work well for the snake robot, but they need to be further investigated and modified.