wps/theme3
Theme 3 : Biommetic EngineeringWP 3.1 : Bat capture, Training and ExperimentationObjectives
Description of workThe proposed project depends on the availability of bats trained toperform their sensory discrimination tasks in situations where their flight behaviour, acoustic emissions and morphology can be observed and measured. This workpackage subserves this need by providing and training as necessary specimens of Myotis daubentonii (in Denmark) and of Macrophyllum macrophyllum, Micronycteris microtis and Noctilio leporinus (on Barro Colorado Island, Panama). Supervise the building of a flight cage large enough for Noctilio leporinus and with suitable facilities for protecting humidity-sensitive equipment. Locate roosts where specimen individuals can be obtained, andcapture individual bats as needed. Prepare an experimental setup that will allow the collection ofthe bat vocalisations during execution of natural behaviours. Train the bats to perform the behaviours of interest in ways thatmake collection of acoustic data feasible, e.g. train the bats to follow particular flight paths where microphones are to be placed. Since the 3D scanner can only capture data from a c.nbsp;30 cm squarecorridor of space and at a camera-fixed maximal frame rate, it is important that the bats are trained to fly along this corridor when being recorded. Train the large Noctilio leporinus to carry microphones in flight. Deliverables
WP 3.2 : Modelling of Relevant Acoustic Targets and Acoustic BehavioursObjectives
Description of workWhen hunting, bats vary their emitted sounds depending on the phase ofthe hunting behaviour (for example, search or pursuit) and on what information previous echoes have yielded. Experiments using naturally behaving bats will be conducted to determinetheir task-related acoustic output and its variability. Existing recorded data will be reanalysed and used where possible. Using reconstructed emitted calls from these naturally behaving bats,and artificial calls with similar parameter choices, echoes will be collected from natural target and clutter objects in ecologically valid situations. These echoes will be examined for acoustic features relevant to the bat's task, using ethological knowledge of bat acoustic behaviour, suitable signal processing and machine learning methods. Once salient acoustic features are identified, signal processing to recover them from cochlear output will be developed. Finally, the variation in call parameters with task phase andpreviously detected acoustic features will be examined so that the dependencies between emission and situation can be elucidated. Computational models capturing this knowledge will be constructed. Deliverables
WP 3.3 : Modelling of Relevant Motor BehavioursObjectives
Description of workWhen hunting, bats use different flight behaviours and vary their headmorphology depending on task needs. For instance, Micronycteris microtis uses a search phase, a scan phase and a grab phase when hunting for insects in clutter; the trawling bats also have distinct search and pursuit phases. Experiments using suitably trained bats will be conducted tocharacterise the distinct phases of their hunting behaviours and identify relevant switching cues. Existing recorded data will also be used where possible. Computational models that generate the observed flight and phaseswitching behaviour in response to echo information and current task phase will be developed. Deliverables
WP 3.4 : Implementation of Biomimetic ModelsObjectivesTwo working robotic implementations, one a model based on Micronycteris microtis attached to a robot arm; the other a model based on a trawling bat attached to a mobile robot.
Description of workThe computational models of acoustic and motor behavioursdevised in WP 3.2 and 3.3 will be integrated with the plastic heads produced in WP 1.1 and a robot base. Infrastructure work involves maintenance of CIRCE head and mobile/armrobotic hardware, fitting of custom emitter/receiver structures supporting dynamic beam-forming to the platforms, and constructing common tools for acoustic and motor control as appropriate. Finally, the computational models of behaviour are realised and testedon their hardware platforms. Deliverables
WP 3.5 : Characterisation and Evaluation of Biomimetic ModelsObjectives
Description of workThe implemented computational models from WP 3.4 are characterised and evaluated. The performance of each model is measured, and the model's robustnesswith respect to task and environment variations is assessed. For the model of Micronycteris microtis, this will include the model's ability to cope with varying leaf density and clutter type as well as different insect targets to pick. For the trawling bat model, it will include variation in water surface smoothness (e.g. wave disturbances) and varying target insect and clutter types (e.g. leaves, sticks) and robustness to `jamming' caused by multiple co-located sensors working independently; additionally for this model, if time permits, it will include inter-species comparisons between Myotis daubentonii and Noctilio leporinus. The models will also be evaluated by comparing their performance tothat of the bats they model, with respect to ability to discriminate targets and cope with variation of conditions. Deliverables
Created by: admin last modification: Monday 16 of February, 2009 [09:23:16 UTC] by admin |
Useful linksA presentation of the main points of the project. Related Projects The CIRCE robot bat head 2002-2005: CIRCE (Chiroptera Inspired Robotic CEphaloid; IST-2001-35144) which reproduced, at a functional level, the echolocation system of bats by constructing a bionic bat head that was used to systematically investigate how the world is not just perceived but actively explored by bats. This bionic bat head is of similar size to a real bat head to reproduce the relevant physics and consist of an emission/reception system capable of generating/processing bat vocalisations in real-time, a multi-degree of freedom mechanical system to allow realistic pinnae movement. 2005-2010: CILIA (Customized Intelligent Life-inspired Arrays; IST-2005-016039), about sensory systems based on arrays of hairs. The project aims to identify the common principles underlying the widespread use in nature of arrays of mechanical sensory cells for the extraction of meaning under adverse conditions and to make those principles available for design of engineered systems. |
|||
