Hexapod Robot - Advanced Digital Projects Lab | ECE 395, Study Guides, Projects, Research of Electrical and Electronics Engineering

Download Hexapod Robot - Advanced Digital Projects Lab | ECE 395 and more Study Guides, Projects, Research Electrical and Electronics Engineering in PDF only on Docsity!

Advanced Digital Project Laboratory

Hexapod Robot

Name : Sanghun Choi

Pedro Nariyoshi

Manisha Choudhury

Object

Since most military vehicles are based on the wheel system and soldiers have serious problems in transporting military equipments and other necessities in mountain areas. Soldiers have to move them on their own, because the normal wheel vehicle cannot be used in mountain areas. If the military vehicle has the capability of moving around, we can be faster in transporting required material in demanded area. But, there was another problem in deciding the right number of legs of the robot. Six has been recommended as a proper numbers of legs because the six leg system has some benefits on designing a vehicle. Since the hexapod robot can be statically stable on three or more legs, it has great deal of flexibility in how it can move. If legs are disabled, the robot may still walk. Robots having six legs or more can be categorized into three different types on the basis of their walking motion. The 3 categories are Wave Leg Motion, Tripod Gate and Quadruped Gate. The Wave Leg Motion is designed to move each of the six legs in sequence in an interval. Since the remaining five legs are able to sustain the main body, when one of the legs try to move, this leg motion is the safest among the three ways of leg motion. But the main process must control six legs differently and give intervals between each legs’ move, thus this leg motion is the most complex and slow. The Tripod gate and the Quadruped Gate are designed to move three or four legs at the same time, and theses two gates are simple and easy to implement the leg motion in. Therefore, we designed the small hexapod robot with Tripod Gate. In addition, when the robot is to be worked in automation mode, the A.I system is needed to avoid obstacles and find the commanded destination. By installing Ultra Sonic Sensor to detect the distance between robot and obstacles, an automatic obstacles avoiding system has been obtained. Finally, the voice recognition system is installed on the robot to follow the commands given by a human being. As the voice recognition system is still in the development stage, one can easily give commands to the robot without any control instructions. Therefore, this project has been focused on developing autonomous robot which can walk with six legs and can recognize human language.

Parts

Robot Body

Commercial Robot Hobby Kit with 13 servo motors from http://www.lynxmotion.com

Servo Motors

6 : Hitec – HS 422 with 4.1 kg.cm Maximum Torque for Horizontal Force

the converted input through HM2007 is to be sent to Command selector by R.F transmitter and receiver system.

RWS-434 data R.F receiver

FRONT

BAC

CIP-8D : R.F 8bit decoder

TWS-434 data R.F transmitter

FRONT

BACK

CIP-8E: R.F 8bit encoder

Ultra Sonic Sensor

SRF04–Ultra Sonic Ranger As shown in the Timing Diagram, the output signal varies based on the distance measured from the obstacles. For example, if the distance is longer, and the time for the ultra sonic signal to be received back after reflecting from the obstacle will increase and vice versa. The timing circuit will then send a pulse back to the microcontroller which is wider for greater distances. The timing circuit is activated when a the module receives an input, so we can adjust how fast we want to detect distances, and thus finding a compromise between energy efficiency and accuracy.

Power

5 Vdc regulator 7805A 7.2 Vdc Battery 6V Vdc 4*AA battery.

Block Diagram for R.F transmitter and receiver

Servo Motor

Servo Motor is based on the PWM (Pulse Width Module-1500usec NEUTRAL), and the motors consists of three external ports. One is Vdd Voltage(Red Line), the other one is Ground(Black Line), and the black wire is finally connected to the signal input. Depending on the width of the pulses received in the input wire the motor will move to a certain direction. For example, when the input signal has a 0.9mS width in a 50Hz signal (20mS period) the motor will turn 45o^ counter-clockwise, in the same way a 2.1mS width pulse will cause a 45o^ rotation clockwise. The 1.5mS will cause the motor to go to the central posittion. Pulse width and servo angle of rotation

Servo Motor Controlling

First of all, generating the pulse width signal with 20ms period is needed to control the servo. We used timer-counter which add and increase one step every 0.1ms. To be able to create simultaneous movement in various legs at the same time (with different phases) and maintain the 0.1mS difference, we programmed a timer function that will change it's state every 0.1mS. And then, to get the desired the pulse width, the sub function is set to be ‘1’ until the counter number reaches desired position. For example, to produce 0.9ms pulse width signal, the sub

implement perfect Tripod gate, which is better than ever, after understanding and physical analysis of the robot. Here is some leg motion sequence of Forward, Turn left, Turn right, Backward. < G1 : Group1 / G2 : Group2 > Forward: G1 (Forward) –G2(Forward)- (Forward) –G2(Forward)-……. Turn left: G1 (Backward) –G2(Forward)- (Backward) –G2(Forward)- Turn right: G1 (Forward) –G2(Backward)- (Forward) –G2(Forward)- Backward: G1 (Backward) –G2(Backward)- (Backward) –G2(Backward The Best Position For Leg’s Downing Position Forward < : Downed State> Fro nt Fr ont

Backward Left and Right Motion will be designed with combination of Forward and Backward each sides.

Avoiding Obstacles

To design an Obstacles Avoiding Autonomous System, we used the ultra sonic sensors, which generate an input signal by measuring distances between obstacles and body, and produces proper command based on the input signal from the Ultra Sonic Sensor. To activate the sensor, a 5Vdc and a pulse integral input signal is needed. With the help of the DSPIC30F4013 micro processor, it is coded to generate the pulse signal with a timer 0 interrupt like PIC16F877A. When the output from the sensor comes to the DSPIC processor, it will count every 0.1ms when the input is ‘1’. Thus it converts the length between obstacles and the robot to binary code. Through distance measuring process, this robot can find other path where there are no obstacles. Finding the Proper path is summarized below in the Block Diagram. Fr ont Fr ont

Voice Recognition System

HM 2007(main processor) The HM2007 is a CMOS voice recognition LSI(Large Scale Integration) circuit. The chip contains an analog front end, voice analysis, regulation, and system control functions. Features: Single chip voice recognition CMOS LSI Speaker dependent External RAM support Maximum 40 word recognition (.96 second) Circuit Diagram

Sensors

IR Sensors

The first kind of sensor we tried was the IR (infrared) sensor. Their principle is very simple, we switch an IR LED at a given frequency (38kHz), the IR wave would then reflect at a surface a come back to our IR receiver (TSOP1738) and the receiver would send a TTL-level signal to our microcontroller. This design is very simple and was easy to build, nevertheless it had big shortcomings. Initial Schematic The first one was the range, the range was very short. We wanted a range of approximately 40 cm so that we could know how far the robot was from an obstacle and have some time before it was too close, so the microcontroller could take action. Several modifications on the original schematic were made to we could a longer range. The initial range was about 5 cm. Raising the voltage to 7.4V at the LED part of the circuit improved the response significantly, but it was still about only 10cm. Raising voltage beyond this was neither practical (We would not have a voltage source beyond that, except for 14.8V (The 2 batteries in series), which was also too high. And also, we found out that beyond that voltage the LED tended to burn, unless we raised the series resistance, which would make the current the same, rendering the same results as a lower voltage and lower resistance. The second measure was to add another LED in series. For that we used an even lower valued resistor, but measured the voltage across it with an oscilloscope so we could know how much current were we putting through the LEDs, also we raised the duty cycle to 80%. Again, we added about 10 cm to the range, but depending on the surface, the range was still shorter than 20cm. Our third solution was to tweak the voltage that powered our receiver. We found out that for lower voltages (4.5V) the receiver was more sensitive and we got to a total of 25 cm range. Which was still not good enough, but we were getting close. But another factor made us change our mind. The power consumption. One set of 2 LEDs and the sensor were consuming about 120mA, since we wanted at least 4 of these sets, the

Mechanics

Servo Motors

We used 2 different models of servo motors, HS422 and HS625MG, both of them manufactured by HiTech. The motors are rated for 6V, but properly filtered, they could handle even 12V (as used in the Sliding Whistle project). The HS422 and HS625MG are exchangeable, they both respond to the inputs in the same way, the HS625MG being much stronger (and more costly). We have had several problems with them, the biggest being torque/speed and energy consumption. At first all of our motors were HS422, which are strong for some applications, but not enough to lift our robot. We then tried several solutions, first, to remove everything from the robot that we did not need to make it lighter. Also we considered making our robot carry a cart with all the boards and batteries, but this would not be neither practical or graceful. Also, the current peaks combined with the inductance produced voltage peaks, which often killed the microcontroller or even the other motors (2 motors died due to voltage peaks), we decoupled the microcontroller and improved the power supply to prevent this, but we also tried to reduce the current peaks. We tried then putting springs in the legs to make it easier for them to move, and the results were promising, we reduced the standing position consumption by around 40%, but we later found out that there was a 40% consumption in the dynamic movement, and since our robot was designed to move rather than stay still, we removed the springs. Also, we tried different position, we discovered that the way the robot moved greatly influenced the current. For instance, if the standing position was the Position A, we could improve in 80% the energy efficiency while moving and standing, when compared to position B. Also we don't need to move the legs all the way up to make it walk, small movements are enough to lift up the robot. Besides that, instead of moving the motor from one extreme to another, we made gradual changes, so the current peaks would be reduced. The vertical-movement motors were replaced by the HS625, because it offered higher torques and higher efficiency. These modifications made our robot move faster and move naturally, even though we can't say that it moves perfectly naturally yet.

Power Supply

As mentioned above, since our motors caused current peaks (and both high and low voltage peaks), we had to design a better power supply. Also, batteries that can provide high currents are usually heavy (as Pb-H 2 SO 4 ), which would cause our robot to be even slower and to require even more power. We decided then to use LiPoly (Lithium Polymer – more specifically LiFePO 4 ). These batteries are used in cellphones and laptops, they are light and rechargeable. The two problems with these batteries is that they cannot provide high currents or fast current variations and they only come in voltages that are multiple of 3.7V (nominal, will vary from 3.25 and 3.8V). Our solution was to buy 4 cells and used them in series obtaining 14.8V and using a step down switching voltage regulator (PTN78020WAH – Manufactured by Texas Instruments), these modules can provide up to 6A continuously and have high efficiency (more than 90%), to provide for the fast current variations we put 2 x 2.2uF monolithic ceramic capacitors in the input and a big 3300uF electrolytic capacitor in the output (And other 2.2uF monolithic capacitors close to the motors. The monolithic capacitors are fast (low ESR and low inductance) and small in size. The electrolytic capacitor is slower (high ESR, and inductance), but can store a lot of charge per unit of volume, therefore we used them in parallel as our filter. Since our duty ratio was lower, we could provide 6V for the motors without wearing out our batteries. And due to its high efficiency, no heat sinks were necessary and no heating was shown by the device. The microcontrollers were connected to another supply, a 5V supply (even though our microcontrollers could withstand 7.5V), since they wouldn't consume too much power, we used a much less energy efficient design using a simple regulator (LM7805) and a 9V battery.

Board

The board design was made in Eagle, some features are worth mentioning, for the buses with higher current, besides being wider, we used both sides of the board, so we would have smaller resistance. We also added a switch to be able to turn on and off the debug-LEDs, to save energy. We also added socket slots to put the phone jacks so we could program the PICs without taking them out of the board. Another feature was the use of 2 PICs, a common PIC and a DSPIC, the DSPIC controlled the voice recognition system and the sensors and sent the commands to the PIC, so in this way it acted as the brain, as the PIC controlled the 12 motors and received the signals from the DSPIC, so it acted as our cerebellum. Our plan was to later move into only one chip (even though Dual Core sounds very trendy nowadays), but in this way we could program the movement and the decision making separately, so that was our reason