click below
click below
Normal Size Small Size show me how
Robotics 101
My Robotics 101 course terms
Term | Definition |
---|---|
Intelligent Agent | An entity which observes through sensors and acts upon an environment using actuators. |
The 4 D's | We use robots for missions that are Dirty, Dangerous, Distant and Dull. |
The 3T Architecture | Planner <-> Execution <-> Control and Sensing affects them all. |
Planner | High level mission planner that plans as a response to Execution request. |
Execution | Manages task execution. Ask for a plan from the planner, decomposes into primitive actions for the motors, decides what to do next. Sends primitive actions to motors, and receives success/failure. |
Control | Low level control. Responsible for activation of motors. |
Remote Control | A human operator who sees the robot and the environment from own eyes. |
Teleoperation | A human operator who sees the world from the eyes of the robot. |
Extereoceptive Sensor | A device that is being used to monitor the relationship between the robot kinematics and/or dynamics with its task, surrounding, or the object being manipulated. |
Proprioceptive Sensor | A device that is being used to measure the position, velocity and acceleration of the robot joint and/or end-effector. |
Vision Sensor challenges | Lightning changes Scale Moving from 3D to 2D creates loss of information Noise |
Laser Scanners (LiDar) | Distance sensors that use the same principle as ultrasonic distance sensors, but use light instead of sound. A LiDar sends a laser pulse in a narrow beam towards the object. |
Laser Sensors Pros | Narrow beam width, higher resolution Not effected by temperature or vacuum Greater accuracy and faster response |
Laser Sensors Cons | Costly Limited in range in various applications or not safe in others Relatively heavy Works badly in poor visibility and underwater |
Kalman Filter | Compare where we think we should be, to where our sensors tell us we are, by taking a weighted average of the two, based on their covariance, as the new state. |
Holonomic Robots | Robots that can move in any direction |
Non-Holonomic Robots | Robots that are constrained to certain types of movement |
Feedback Control | Check where located and where should be, calculate the error and take action to minimize the error. |
Goal Achievement | Once reaching a goal state, job is done. |
Goal Maintenance | Ongoing active work of staying in the goal state. |
Bang-bang Control | Simple on/off switch. When error is negative, on When error is positive, off Good for reaching goal, not good for keeping it. |
Proportional (P) Control | Signal becomes proportional to the error. Output as a function of input and constant. o = K_p * i The larger the error, the greater magnitude of response. |
Proportional Derivative (PD) Control | Like P control, but taking into account how fast approaching the setpoint o = K_d(1 + (curr_measure - prev_measure)) Commonly used as low-level controller |
Integral Control | Keep track of steady state errors o = K_f * integ(i(t)dt) |
PID Control | o = K_p * i + K_d * di / dt + K_f * integ(i(t)dt) It's impractical if the process is complex. |
Feedforward Control | Does not use sensor input, but it predicts system state. Works only for well calibrated systems. |
Sample-based planning | Exploring the state space using sample points. The points are acquired by drawing randomly a point in the environment, and connecting it to the closest point(s) in the tree. |
Coverage | Finding a path for the robot(s), such that each point in the area is visited once [in minimal time]. |
Dead Reckoning | The process of calculating one's current position by using a previously determined position or fix, and advancing that position based upon known estimated speeds over elapsed time and course. |
Odometry | The use of data from motion sensors to estimate change in position over time. |
Separation Bearing Control (SBC) | The robot position itself against a single robot |
Separation Separation Control (SSC) | The robot position itself against several robots |
Task Specific Planning Theme | For world representation we use something known, the algorithm takes most of the effort, and if the modeling was correct, path tracking is trivial. |
General Planning Theme | The world representation takes most of the effort, for the algorithm we use something known, and if the modeling was correct, path tracking is trivial. |
BDI Architecture | Methodology for implementation of practical reasoning agents. |
BDI Components | Beliefs - knowledge base; Plan library; Goals - a state(s) of the world, the agent wants to achieve; Intentions - an option (a plan) the agent committed to achieve; Options - plans the agent might perform given its beliefs and its previous intentions |