What has Evolution to say about EmbeddedSystems?1953 Miller−Ureyammonia + H O + H + CH +ENERGY23amino acidsalanine+glycine−4.6B rocks − primordial soup−3.5B single celled organisms−2.5B photosynthetic plants−0.5B fish, vertebrates, insects, reptiles,dinosaurs, mammals−0.002B man1 Copyrightc°2004 by Roderic A. GrupenThe Human Age−18M−2.5M−0.01Mgreat apesmanagriculturewritingexpert knowledgeBrooks, Morevec, et. al....moving about in our complex environment well enough to surviveand flourish is the toughest problem that life on earth has yet faced......this is where evolution has concentrated most of its effort.figuring out how to build “embedded systems” - tocombine computation, control, signal processing,mobility, and embodiment - is where mother naturespent most of her time so far2 Copyrightc°2004 by Roderic A. GrupenEmbedded Systems - Lecture Units artificialintelligence RTcomputingelectronics signaldetectionkinodynamicsmechanismenvironmentembedded systems3 Copyrightc°2004 by Roderic A. GrupenEarly Embedded Control SystemGrey Walter’s Tortoise• 1953 - Product of cybernetic research• One motor with 2 valve control circuit• Two sensors: light and bump• Light sensor always faced in direction of motion• Behaviors– Seek light (rotates until light found)– Head toward weak light (once light found)– Back away from bright light– Turn-and-push to avoid obstacles: avoid obstacle overrodelight response– Recharge battery - strong source perceived as weak light withlow battery, once charged backs away from plug.• From lowest level to highest:scan, move to/from light, avoid-obstacle4 Copyrightc°2004 by Roderic A. GrupenEarly Embedded Control SystemGrey Walter’s Tortoise5 Copyrightc°2004 by Roderic A. GrupenEarly Embedded Control SystemBraitenberg’s Vehicles• Simulation study (1984)• Sensors mapped to motors via simple analog logic• Synthetic psychology — seemingly complex behavior results fromrelatively simple sensorimotor transformations.• Inhibitory and excitatory influences (sensors on motors)• Range of vehicles created - e.g., Coward, Aggressor, Lover, Ex-plorer.• Inflexible custom machines6 Copyrightc°2004 by Roderic A. GrupenExample of Reactive ControlBraitenberg machinescowardexplorerloveraggressor(From Braitenberg 1984)7 Copyrightc°2004 by Roderic A. GrupenExample: Raibert’s Hopping Platformshttp://www.ai.mit.edu/projects/leglab/robots/robots.htmlTHRUSTUNLOADFLIGHTLANDCOMPRESSIONlift−offtouch downleg shortensbottomleg extended8 Copyrightc°2004 by Roderic A. GrupenProject Concept...Marketing has identified a significant market potential for anadvanced robotic toy that can act as a playmate/assistant forchildren and adults.• it must be capable of moving around in human-scale environ-ments(mobility)uBot video...• it will use ... “emoting” to capture user attention and engagehumans in a social context,Leonardo video...•... and interacting manually with the user and the environment.This last issue is the objective of theTorso Workgroup9 Copyrightc°2004 by Roderic A. GrupenTorso Workgroup TaskThe design of the bi-manual “torso” is specified in terms of marketingbehavioral targets, including a representative set of grasping tasksand objects.1. kinodynamics -(a) a trunk rotation of ±60 deg measured relative to the drivewheels(b) the workspace of individual arms must include the floor infront and behind the robot, and must be able to reach verticalconfigurations(c) the bi-manual workspace must include:• a 6 inch deep region on the floor at zero trunk rotation• configurations that can be used to brace for a fall and rightitself after a fall2. function -(a) the device should be able to follow “hand-in-hand” behind ahuman client(b) must be capable of grasping a cylinder 2 inches in diameterand 2.5 inches tall on the floor3. Cosmetic - arms should be as anthropomorphic as is practical10 Copyrightc°2004 by Roderic A.