Through the Institute's BBD^SM research and development programs, we have constructed a series of physical devices with simulated nervous systems that guide behavior. These BBD^TM brain-based devices are not like conventional robots designed by engineering principles. They are in fact intelligent thinking machines. BBD^TM devices are designed based on biological principles and are programmed to alter their behavior through self-learning.

The first BBD^TM device, known as NOMAD (Neurally Organized Mobile Adaptive Device), was developed by Dr. Gerald Edelman, the Founder and Director of the Institute, in the late 1980s. For a recent interview with Dr. Edelman about his latest book Second Nature: Brain Science and Human Nature click here.

NOMAD learns from its experiences using its simulated brain called Darwin, which is based on principles of vertebrate brain organization encoded in a computer program. The BBD^TM Darwin brain is a “selectional system,” not an “instructional system,” according to Dr. Edelman. Robots are computers that carry out unambiguous logical tasks, while the human brain does not work by logic, but by pattern recognition. Through the Institute's BBD^SM research program, several generations of BBD^TM NOMADs and Darwins have evolved into a new and powerful class of intelligent machines.

NOMAD possesses auditory sensors, a video sensor, ultrasonic range sensors, a gripping device which includes electrical conductivity sensors, and two sets of "whiskers" that can feel as the device contacts objects in its environment. It moves on its wheels and can rotate in place. Although some low-level controllers reside in the device, the main control is through a "nervous system" simulated in an array of computers with which the device communicates over a wireless connection. The nervous system integrates the various sensory inputs and responds with motor control actions that are then carried out by the device. NOMAD navigates on wheels through its environment, senses obstacles and avoids bumping into them, approaches objects after viewing them from a distance, grabs objects with its gripper, tastes objects, and avoids objects that it has learned taste bad.

The principal focus of the BBD^SM research at the Institute has been to test theories of the functions of a nervous system embodied in a real world, however our results may also provide basis for practical applications.

Spatial and Episodic Memory
The hippocampus is an area of the brain needed for storing and retrieval of memories. The BBD^TM Darwin X incorporates aspects of the detailed anatomy of the hippocampus and surrounding brain regions (cortical areas for vision, space, and self-movement) known to be necessary for acquisition and recall of spatial and episodic memories. The BBD^TM Darwin X successfully finds its way around a task maze. After a period of exploration it finds hidden, but remembered locations, and moves toward them from any starting place. All of these advancements are applicable in natural settings, an area of research that the Institute is actively pursuing.

Fine Motor Control and Movement
A BBD^TM brain-based device has been constructed which incorporates a detailed model of the cerebellum, an area of the brain needed for accurate motor control. This BBD^TM device was given the task to navigate a path defined by traffic cones. At first navigation was clumsy as it relied on reflex from the infrared proximity sensors that were triggered when the BBD^TM device was within 12 inches of a cone. Over time, as the cerebellar circuit became able to predict correct motor responses based on visual motion cues, the movements of the BBD^TM device became smoother. This fine tuning of motor control may have significant applications for machine control, as well as for understanding human movement.

Soccer-Playing BBD Device
We have created a BBD^TM device especially designed to play a version of soccer using the Segway platform transporter. This BBD^TM device recognizes objects in the field (balls, teammates, goals, etc.), and uses a specially designed mechanical device to capture and kick the ball. Our team of scientists has participated in the US Open RoboCup Championships. In 2005 our BBD^TM soccer-playing device was undefeated, competing against soccer-playing devices designed at Carnegie Mellon University

Download the complete Scientific Report (PDF)

back to top