Modeling General Principles Of Motion : Generating A Species-agnostic Model Of Motor Control (IRG1 - Project Reference 1938759)

We will build a scale-dependent model of how lower-level motor centers (LLMCs..i.e. local ganglia) control rhythmic motions that require interaction with the environment, such as walking or grasping objects, and how higher-level command centers (HLCCs...i.e. central cerebral structures) interface with the LLMCs to modulate behavior. Our overarching hypothesis is that the dynamical scale of a behavior, that is, the size of the animal and the speed of its motion, will enable us to derive cross-phylum principles of motor control and predict what neural circuitry each animal possesses. Our group consists of modelers from within three collaborating individual research groups (IRG's 2-4). We will use neuromechanical models of animals to generate hypotheses to be tested IRG's 2-4, and then compare and generalize what is uncovered by experimentation. We will make the whole research team of 4 individual research groups greater than the sum of its parts, by providing a theoretical framework to guide experimentation and integrate results from the other three research teams. We will model and analyze how the dynamical structure of LLMCs and the body itself simplify what information must be shared between the HLCC and LLMC, by considering a series of questions: How does the size of an animal and the speed of its behavior affect its passive mechanics? Do an animal's passive mechanics change the gain or type of sensory information necessary for control? How do the mechanics of the body simplify the descending control of behavior? The general principles uncovered will be validated by their use in controlling the walking and grasping of a general-purpose robotic platform.

Our modelling and robotics work will be driven by two specific hypotheses:

Hypothesis 1: Quantifying the dynamic scale of an animal and its behavior will enable the prediction of sensory feedback used to control motion. The ratio of elastic (or viscous) to inertial forces increases as animals become smaller (or slower). Small and/or slow animals have less gain of negative feedback reflexes as compared to larger animals, because the elastic and/or viscous forces stabilize behavior.

Hypothesis 2: Despite differences in scale across species, there is a fundamental organization of the LLMCs that stabilizes interactions with the environment. This organization consists of pattern generators and reflexes within the LLMC, as well as motor responses mediated via ascending pathways to the HLCC and descending commands back to the LLMC. The efficacy of such pathways is enhanced by a simple, abstract code for communicating information.

Four specific aims will drive us to test these hypotheses
Specific Aim 1: Refine theories of how scale affects passive body forces.
Specific Aim 2: Construct an initial modeling framework common to C3NS's model organisms
Specific Aim 3: Understand and model organismal commonalities in how descending and ascending information is encoded, transmitted, and decoded by lower and higher-level networks
Specific Aim 4: Validate the generalized model of the control of interaction with the environment.

This grant application is for the development of a general motion control model, taking account the different physical regimes under which animals move. We have developed a quantitative control analysis that brings animal motion control down to two fundamental variables: 1) the time of the behaviour and 2) the size of the animal. This mathematical formulation allows us to propose a control strategy that characterizes the mechanical resonance of the system (its characteristic frequency), and compares that to the time-course of the neural inputs - with 'fast' systems having neural inputs that are much 'faster' than the resonant frequency, 'intermediate' systems having neural inputs that are equivalent in time as the resonant frequency, and 'slow' systems in which neural inputs are much slower than the resonant frequency of the system.

Because a system's resonant frequency is the square root of stiffness divided by mass, this frequency is very much determined by the size of the organism; with both mass and stiffness being size dependent. Consequently, we will be using data from the other individual research groups to look a the control of 'fast', 'intermediate', and 'slow' systems across the size range of mg (flies), g (Aplysia), to 100's of g (mice) to further refine this analysis.

Lastly, we will include viscous damping to the control parameters, which will, in theory, allow our analysis to extend down to the microgram level, thus including locomotory and control strategies down to extremely small organisms. With each iteration of the control theory illustrated in case for support, we will extend the size range and physical accuracy of the model to generate a more accurate, species agnostic, model for control of organisms; with the aim to cover a size range from the extremely small (mg and smaller) to the extremely large (whales).

Our Network's research will have broad impacts in biology, robotics, and health. The Network will directly impact biology: Exploring convergent mechanisms of communication between levels of the nervous system and volitional motor control across phyla and scales will lead to a better understanding of the diversity of life on Earth, and may suggest general organizing principles for the nervous system. Additionally, we will improve the state of the art in robot control by applying the generalized principles uncovered by this research to designing and tuning neural controllers for robots. This research will potentially lead to smarter machines whose motion is reliable and robust to perturbations, increasing their autonomy and mobility for work in challenging terrains such as on farms, mines, and other planets. More reliable control and movement could also allow robots to more safely operate among humans such as in long-term care facilities. Generalized principles may also be applied to artificial intelligence, suggesting organizing principles for neural systems beyond the often-simplified architectures such as Deep Neural Networks. Furthermore, understanding the language of communication between different levels of the nervous system will aid in the investigation of motor disorders that disrupt locomotion and balance. It could also lead to the development of synthetic nervous systems for the control of prostheses and orthoses that can effectively process descending neural signals from the brain, and report sensory feedback back to the brain. The patient could learn to use these devices intuitively, as extensions of their body.

Our Network will also expand and enrich our outreach activities with our research and collaborations. Public demos, daycamps and internships will expose K-12 students to interdisciplinary research and international collaborators' culture. Such activities will be coordinated to coincide with the Network exchange program (See section 3.2 Training in the connected documents) to provide a diverse pool of role models. We will leverage existing STEM bridge programs at member institutions to recruit participants.

Our Network will enable us to conduct two new major Network-wide outreach activities that promise to encourage thousands of K-12 students to consider STEM in higher education. First, we propose to construct exhibits at natural science museums in our Network's major cities. This is possible because most of the PIs have contacts and previous collaborations with local museums. We will develop an exhibit that will be copied and displayed concurrently at local museums. The exhibit will include QR-code tags to lead the visitor to our second major initiative, an interactive website. Both of these efforts will describe how different animals solve similar problems in scale-dependent ways with an emphasis on how animals volitionally control their bodies. We will tie the museum exhibits to our daycamps where possible. We will have students attend the museum exhibit prior to the daycamp to learn how biology, neuroscience, and robotics can interact, thereby allowing students to develop creative solutions while in attendance at subsequent camps.

In order to further disseminate the outcomes of our NeuroNex-funded projects, we will encourage Network members to take part in FameLab and other public-outreach events to inform the public about our efforts. These events are usually recorded, and the videos shared online. We will build a collection of publicly available videos in which our grad students explain our efforts in short, easy to understand videos.