Introduction

The primary requirements for the robot arm is that it must have two degrees of freedom, be able to produce angles with a relatively high degree of accuracy and there must be a system in place for reading the arms coordinates or position in space. Keeping the arm simple, small and lightweight were just some of the secondary conditions for this portion of the project to initialize itself.

Parts of the robotic arm

The arm is initially set up to have two degree of freedom with an upper and lower arm segments and two joints. The arm segments are the simplest portion of the robot arm, arm lengths are not necessarily a concern and does not have to be a specific length as will be presented in the learning system portion of the overall project done by the cerebellum.

A requirement of the joints is that they must produce given angles with relative accuracy. Each of the pivoting joints of the arm can be considered the shoulder and the lower joint the elbow. There are many possible setups that can produce this effect. There are hydraulics, chain driven, motors and more specifically servos. For the initial scope of this project the HS-645MG servos from Lynxmotion were chosen. The reason that servos were chosen is that they are light, cheap, easy to use and do not require a lot of customizing.

Interfacing with the robotic arm

Initially the interfacing with the robotic arm will be done using LabView. There are several layers that exist to interface the servos with the LabView software. If LabView running on a Windows PC is the first layer then the National Instruments PCI DAQ card model PCI-6723 is the second layer. From the PCI card runs a data cable to the third layer which is a National Instruments DAQ pad model SCC-68 from National Instruments. From the two analog output ports 22 and 21 are the two signals that run to the servos, the fourth layer. The robotic arm gives back no data feedback as to it’s condition which makes it the final layer in the interface.

Feedback System

The robot arm on it’s own does not have the ability to determine its own position in space. The system, however, requires knowing the arms current position of the arm to correct for inaccuracies in the arm segment lengths. Having the actual and desired positions will be processed by the brain to be used for weighting. Much like deciding on how to implement a joint the feedback device has a lot of possible choices. These choices include sensors(sonar, infrared, touch pads and feelers) and image processing. Image processing was chosen because it is becoming a more widely accepted method for acquiring data for robots and feels more natural than other more traditional methods like using sensors. More data can be acquired on the Visual Detection System project page.

Data Specifications

The servos were calculated in various experiments. In Table 1 below is the results of the experiments preformed. Clicking on the header of each of the specifications will take you to the write ups for those values. During the June 1, 2007 experiment a standard for the servo angles was established using servo number 2 which creating a set of equations that can be used to convert from angles or pulse width times to a PWM signal.

Experiments

Lab Report: Jesse June 1, 2007
Lab Report: Jesse June 8, 2007
Lab Report: Jesse June 14, 2007
Lab Report: Jesse June 22, 2007
Lab Report: Jesse June 29, 2007
Lab Report: Jesse July 6, 2007