Title
Connecting Brains to Robots: The Development of a Hybrid System for the Study of Learning in Neural Tissues

Abstract
We have developed a hybrid neuro-robotic system based on a two-way communication between the brain of a lamprey and a small mobile robot. The purpose of this system is to offer a new paradigm for investigating the behavioral, computational and neurobiological mechanisms of sensory motor learning in a unified context. The mobile robot acts as an artificial body that delivers sensory information to the neural tissue and receives command signals from it. The sensory information encodes the intensity of light generated by a fixed source. The closed-loop interaction between brain and robot generates autonomous behaviors whose features are strictly related to the structure and operation of the neural preparation. In this paper we provide a detailed description of the hybrid system and we present experimental findings on its performance. In particular, we found (a) that the hybrid system generates stable behaviors; (b) that different preparation display different but systematic responses to the presentation of an optical stimulus and (c) that alteration of the sensory input lead to short and long term adaptive changes in the robot responses. The comparison of the behaviors generated by the lamprey's brainstem with the behaviors generated by network models of the same neural system provides us with a new tool for investigating the computational properties of synaptic plasticity.

Source: Artificial Life VI (2000)

Title
Connecting Brains to Robots: an Artificial Animal for the Study of Learning in Vertebrate Nervous Systems

Abstract
We have developed an artificial animal, consisting of the brain of a lamprey (a primitive vertebrate) controlling a small mobile robot. The mobile robot acts as an artificial body, which delivers sensory information to the brain through its light sensors, and is controlled by command signals generated by the brain itself. As the behaviors resulting from the interaction with the external environment are specified by the structure and operation of the neural circuitry, this system provides a unified context for investigating the behavioral, computational and neurobiological mechanisms underlying sensorimotor adaptation. In this paper, we provide a detailed description of the system and report on a number of experiments in which we have analyzed the behavior of the animal in response to a light stimulus. Most preparations generated stable and repeatable behaviors; different preparations showed different behaviors, with a slight preference for moving toward the light source. The observed behaviors were consistent with those predicted by the measured input/output responses of the neural circuitry. In addition, we observed long-term behavioral changes taking place in response to repeated tonic optical stimulation on one side of the animal. Our results suggest that the system may be used as a new experimental paradigm for investigating the computational properties of synaptic plasticity in the context of sensorimotor adaptation.

Source: Animals to Animats 6 (SAB 2000)

Title
A Neuro-Robotic Interface for the Study of Synaptic Plasticity in Sensorimotor Adaptation

Abstract
This dissertation details my investigation of synaptic plasticity in the Sea Lamprey Petromyzon marinus within the context of sensorimotor adaptation. This research, derived from two studies, one using extra- and intracellular recordings of an in vitro lamprey neural preparation and a second using a novel neuro-robotic interface. The neuro-robotic interface allows for the simultaneous investigation of cellular and behavioral changes associated with synaptic plasticity.

The first study tested the hypothesis that the intensity of a low frequency train of stimuli applied to vestibular axons is correlated to both the magnitude and direction (potentiation or depression) of synaptic plasticity observed in the vestibulo-reticular synapses. The degree of association, an indication of the number of afferent fibers recruited and dependent on the level of stimulation, was found to be linearly correlated with the level of plasticity found in the vestibulo-reticular synapses. This finding provides a physiological explanation for the quick recovery of certain animals after impairment of their vestibular system.

The second investigates cellular and behavioral changes in the context of sensorimotor adaptation. A hybrid system was developed that interconnects a lamprey brainstem preparation, maintained in vitro, to a mobile robot through a neuro-robotic interface. Therefore, the mobile robot plays the role of an "artificial body." The interface transforms information from robot sensors into artificial vestibular signals, and from reticular neural activity into robot motor commands. The system is capable of generating stable artificial behaviors for extended periods of time (8-10 hours). Both short and long-term plastic changes, due to modifications to the lamprey's sensory system, were observed. A technique to measure the state of neural connectivity of the lamprey's vestibulo-reticular system was developed. A linear model was used for the vestibulo-reticular neural network to develop a catalog of possible robot behaviors generated by the hybrid system. These behaviors range from light following, phototaxis, to light circling, menotaxis. With these findings one can compare, qualitatively, and evaluate the changes in artificial behavior generated by the hybrid system as changes in the neural system.

Source: Dissertation, Northwestern University, September 2000

Title
Templeton: Design of a Robotic Rodent

Abstract
A robotic rat was designed and built to assist in testing of the vector field control algorithm proposed by Mussa-Ivaldi, et. al. The robot, named Templeton, was built to closely follow the behavior of a real rat. Relationships that guide the proper scaling of the robot were derived. The final size of the robot is approximately 3.5 times larger than an ordinary common rat (rattus rattus), but its dynamic properties are similar to that of the real rat. Templeton’s hip is a three degree of freedom joint with angle sensing on all three axes. Both the knee and ankle joints are one degree of freedom joints with similar angle sensing equipment. Templeton’s muscles are custom built pneumatic actuators, designed to mimic the twitch response of biological muscles. The actuators are of a glass/graphite construction that results in low friction. They are controlled by electric solenoids which in turn are a operated by computer. Tendons on the robot are stainless steel cables attached at one end to a link and the other to an actuator. In line with the tendons are strain-gauge sensors designed to measure tension.

Source: Masters Thesis, Northwestern University, September 1996

Title
Robotic Rodent : Progress Report

Abstract
Since April I have been actively invlolved in the design of a miniature robot. The robot that I was to create was a rat, and it was to be as close to scale as possible. In this report I will define rat scale as claiming that the femur and tibia/fibula of the hind leg were each approximately 4 cm. long.

Source: Northwestern University, July 1995

Title
Second Generation Robotic Rodent Actuator

Abstract
The goal of this project is to creat an actuator for a robotic device on the scale of a rodent. The actuator must be small enough to fit on the skeletal structure of a rodent as well as be able to create the sufficient force to generate movement. This report discusses the design and prototyping processes that were involved in creating the beta version of the actuator. The design modifications are also included.

Source: Senior Design Project, Northwestern University, Spring 1995

Title
Micro Actuator Design For a Robotic Rodent

Abstract
The goal for this project was to create an actuating device that would fit the needs and condition restraints imposed. The actuator is to be ultimately used in a robotic rodent. In order to achieve this goal a feasible and practical design had to be created. This design must be followed in the manufacturing of the device. Once manufactured the performance of the device must be tested.

The design and fabrication of the device was completed however the testing scheme was plagued with misfortune. Definitive results were unable to be taken from the device. Estimations from visual data yielded frequency and amplitude estimations.

Source: Northwestern University, August 1994