Biomimetic Autonomous Attitude Control

Acronym: 
BIONA
Research Areas: 
A
D
Abstract: 

We are interested in exploring biomimetic guidance and navigation strategies and applying them to autonomous robots. BIONA is particularly focused on investigating stabilization techniques used by flying insects, and replicating these in suitable small- or micro-aerial vehicles. To this end, we are developing robotic visual systems which, in their physical characteristics and internal connections, mimic the sensory capabilities of insect eyes.

 

Methods and Research Questions: 

How can we design simple, streamlined systems which are able to navigate expertly through a complex three dimensional world? By investigating the behavior, capabilities and neural organization of insects, we can design efficient control mechanisms which take advantage of millions of years of selective evolution.

Many insects lack a reliable gyroscope, yet can maintain stabilized flight or conduct extreme aerobatic maneuvers without difficulty. BIONA researches the links between attitude stabilization and vision in insects by building models of the neural pathways inside the insect brain, and examining the behavior of bees and other flying animals in navigational experiments. We are interested in the performance of both the compound eyes and the secondary eyes (ocelli), as these highly interlinked systems are jointly implicated in the head and body stabilization of various insects.

Ultimately we seek to develop an integrated adaptive visual attitude control system which can take advantage of insect-inspired characteristics such as the dorsal light response, spectral sensitivity, biologically plausible spatial and temporal resolution, polarization vision, neuronal pooling and multimodal interaction.

Using existing and ongoing research into the neurological systems of flying insects, and combining this with the results of behavioral experiments in naturalistic settings, we are constructing holistic models of flight control in the natural world. This leads directly into the simulation and modeling of complex biomimetic controllers, by taking the features of the vision systems of different insects (primarily bees, dragonflies, flies and locusts) which are likely to be most relevant and useful when controlling a flying microbot, and weaving them into a complex multimodal visual controller (this modelling is chiefly conducted in Matlab). Simultaneously, we are constructing biomimetic sensors incorporating panoramic multimodal vision systems such as that developed in the concurrent project FLINAVS. Using this parallel development, the controller design and sensor design are fully integrated as an embodied system. Later this year we will conduct model verification tests and experiments on ground-based autonomous and/or semi-autonomous robots, as a precursor to implementation on a flying vehicle.

 

Outcomes: 

The expected outcome for BIONA is the development of an attitude stabilization system which can be integrated with other visual navigational systems and implemented on a light-weight autonomous flying vehicle, such as a quadrocopter. This will incorporate bio-inspired sensors and control mechanisms, which extract attitude from a similar visual panorama as that available to insects. As a further benefit, we hope to obtain insight into the control mechanisms used by flying insects in a complex environment. This dovetails with the projected results of the FLINAVS project, which seeks to establish the basis for visual navigation in bees via experiments and modeling.

 

Publications: