ROBOTICS 01PEEQW Basilio Bona DAUIN Politecnico di Torino Mobile - PowerPoint PPT Presentation

ROBOTICS 01PEEQW Basilio Bona DAUIN – Politecnico di Torino

Mobile & Service Robotics Supervision and control

Traditional approach Traditional artificial intelligence considers robot “brains” as sequential processing units Main ideas � Representation -> reasoning -> planning � Model building (maps of the environment is required) � Functional decomposition; hierarchical systems � Symbolic/semantic manipulation Basilio Bona - DAUIN - PoliTo 3 ROBOTICS 01PEEQW - 2015/2016

Supervision and Control a priori knowledge Task/mission commands Position Global map Path planning Localization & Map Building & Reasoning Data Data Motion control Perception treatment treatment iteration commands data Sensors Actuators Environment Basilio Bona - DAUIN - PoliTo 4 ROBOTICS 01PEEQW - 2015/2016

Supervision and Control Position Path planning Global map Localization Reasoning Map Building Local map Path World model Motion Perception control Environment Basilio Bona - DAUIN - PoliTo 5 ROBOTICS 01PEEQW - 2015/2016

Control Strategies � Control loop requirements � World changes dynamically � A compact model of the world is very difficult to define � There are many sources of uncertainty, both in the world and in the robot � Two possible approaches + a combination of them � Classic AI – deliberative model � Approximate world modeling (model-based method) � Based on a set of functions � Vertical decomposition � Top-down approach � Modern AI – reactive model � No world model is given: behavior-based control � Horizontal decomposition � Bottom-up approach Basilio Bona - DAUIN - PoliTo 6 ROBOTICS 01PEEQW - 2015/2016

Control Characteristics Sense – Plan – Act Subsumption/Reactive model This architecture may produce a slow and delayed response Function 1 sense Horizontalapproach Vertical approach Function 2 use model Function 3 plan Function 4 act Function 5 Basilio Bona - DAUIN - PoliTo 7 ROBOTICS 01PEEQW - 2015/2016

Model-Based Approach � Requires the complete modeling of the world � Each block is a computed function � Vertical decomposition and nested-embodiment of functions An example sensors Perception Localization - Map building Cognitive planning Motion control actuators Basilio Bona - DAUIN - PoliTo 8 ROBOTICS 01PEEQW - 2015/2016

Model-Based Approach Planner GOAL RECOGNITION GLOBAL PATH PLANNING Navigator SUB-GOAL FORMULATION LOCAL PATH PLANNING Pilot TARGET GENERATOR DYNAMIC PATH PLANNING Path monitor TARGET LOCATION PATH CORRECTION/OBSTACLE AVOIDANCE Controller COMMANDS Low level control SENSORS ACTUATORS Basilio Bona - DAUIN - PoliTo 9 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Basilio Bona - DAUIN - PoliTo 10 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Rodney Brooks, inventor of this approach, writes: � Complex behavior needs not necessarily be the product of a complex control system � The world is its best model � Simplicity is a virtue � Intelligence is in the eye of the observer � Robots should be cheap � Robustness in the presence of noisy or failing sensors is a design goal � Planning is just a way of avoiding figuring out what to do next � All onboard computation is important � Systems should be built incrementally � No representation. No calibration. No complex computers. No high band communication Basilio Bona - DAUIN - PoliTo 11 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Various names given to this approach � Subsumption architecture � Reactive system � Reflexive behavior � Perception-action Action 2 Perception 2 Perception 1 Action 1 WORLD Basilio Bona - DAUIN - PoliTo 12 ROBOTICS 01PEEQW - 2015/2016

Subsumption architecture � The subsumption architecture was originally proposed by Brooks [1986] � The subsumption architecture copies the synergy between sensation and actuation in lower animals such as insects � Brooks argues that instead of building complex agents in simple worlds, we should follow the evolutionary path and start building simple agents in the real, complex and unpredictable world � From this argument, a number of key features of subsumption result Basilio Bona - DAUIN - PoliTo 13 ROBOTICS 01PEEQW - 2015/2016

Subsumption architecture 1. No explicit knowledge representation is used. Brooks often refers to this as “ The world is its own best model ” 2. Behavior is distributed rather than centralized 3. Response to stimuli is reflexive – the perception-action sequence is not modulated by cognitive deliberation 4. The agents are organized in a bottom-up fashion. Thus, complex behaviors are fashioned from the combination of simpler, underlying ones 5. Individual agents are inexpensive, allowing a domain to be populated by many simple agents rather than a few complex ones. These simple agents individually consume little resources (such as power) and are expendable, making the investment in each agent minimal 6. Several extensions have been proposed to pure reactive subsumption systems. These extensions are known as behavior- based architectures Basilio Bona - DAUIN - PoliTo 14 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Characteristics � No model is necessary � Horizontal decomposition � Coordination + Priority = Fusion � Biomimesis = observe and copy animal behavior � Subsumption � Embodiment Avoid obstacles Discover new areas ∑ Detect goal position sensors actuators Communicate data Recharge Follow right/left wall Coordination / fusion Basilio Bona - DAUIN - PoliTo 15 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Basilio Bona - DAUIN - PoliTo 16 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Basilio Bona - DAUIN - PoliTo 17 ROBOTICS 01PEEQW - 2015/2016

Behavior-Based Approach Basilio Bona - DAUIN - PoliTo 18 ROBOTICS 01PEEQW - 2015/2016

Subsumption architecture Basilio Bona - DAUIN - PoliTo 19 ROBOTICS 01PEEQW - 2015/2016

Subsumption architecture Basilio Bona - DAUIN - PoliTo 20 ROBOTICS 01PEEQW - 2015/2016

Subsumption architecture Basilio Bona - DAUIN - PoliTo 21 ROBOTICS 01PEEQW - 2015/2016

Control Strategies: deliberative vs reactive DELIBERATIVE REACTIVE Model-based Behavior-based Purely symbolic Reflexive Speed of response Predictive capabilities Depends on accurate world models • Depends on the world • Representation-free representation • Real-time response • Slow response • Low level intelligence (stimulus- • High level intelligence (cognition) response) • Variable latency • Fast and easy computation Basilio Bona - DAUIN - PoliTo 22 ROBOTICS 01PEEQW - 2015/2016

Embodiment � To embody (verb) = to manifest or personify in concrete form; to incarnate; to incorporate, to unite into one body � Embodiment is the way in which human (or any other animal) psychology arises from the brain & body physiology � Embodiment theory was introduced into AI by Rodney Brooks in the ‘80s. Brooks have claimed that all autonomous agents need to be both embodied and situated � The theory states that intelligent behavior emerges from the interplay between brain, body and world. Brain, body and world are equally important factors in the explanation of how particular intelligent behaviors originate in practice � Brooks showed that robots could be more effective if they “thought” (planned or processed) as little as possible � The robot's intelligence is organized only for handling the minimal amount of information necessary to make its behavior be appropriate and/or as desired by its creator Basilio Bona - DAUIN - PoliTo 23 ROBOTICS 01PEEQW - 2015/2016

Embodiment � Rolf Pfeifer (AILab Zurich) says that there are essentially two directions in artificial intelligence: � one focused on developing useful algorithms or robots; � and another direction that focuses on understanding intelligence, biological or artificial � In order to make progress on the second direction, an embodied perspective is mandatory Basilio Bona - DAUIN - PoliTo 24 ROBOTICS 01PEEQW - 2015/2016

Situated robot � A situated robot is one which does not deal with abstract representations of the world (which may be simulated or real), but rather reacts directly to its environment as seen through its sensors � An alternative to having a situated robot would be one which builds up a representation of its world and then makes plans based on this representation � Because of the limitations of our present technology, these two approaches often seem contradictory � At present, each approach can be appropriate for different applications Basilio Bona - DAUIN - PoliTo 25 ROBOTICS 01PEEQW - 2015/2016