Mihotel

Mihotel Project Document.

GitHub commit activity GitHub issues GitHub closed issues

📊 View our ZenHub Workspace

🌏 View our patio interactively (it may take seconds to load the model)

🎥 View our simulations

Table of Contents

✔️ Highlights

⚠️ Precautions

🐛 Known Problems

Configurations

List of tools, modules with their version

Item Version Notes
Simulation Webots R2020b we are using a very new version of Webots Nightly Build (24-4-2020)
Python >3.6  
numpy 1.17 Numpy module for python
opencv-contrib-python 4.2.0.32 OpenCV module for python

🔍 Output Description

Style prefix Description
Bright Green [Info]  
Bright Red [Debug] debug information, the difference against info is that, this should not show up unless is debugging
Bright Blue [Command] command given to chassis
Bright Yellow [Detected] detect of object

Solution

System

Structure

The system sets up 3 child processes, one for chassis controlling, one for decision, one for detection.

gray boxes are shared variables between processes

Communication

Queue

Four queues, signal_queue, command_queue, sensors_queue, motors_queue are used for communications between processes. Although it seems when there is only two endpoints to communicate, Pipe() is a faster choice, but it seems the code could be prettier with Queue().

❗️ notice that once Queue.get() is used, one item in the queue is taken out and returned, which means it is not in the queue anymore and you could not get it again with Queue.get(). Queue.empty() could be used to detect whether it is empty.

Shared variables

A few shared variables are created to share some flags and signals between processes. So far flag_patio_finished, flag_pause, key are used.

📚 document for multiprocessing.Value()

Here is a list of one character typecode can be used in multiprocessing.Value() to determine type of the shared variable.

How it ends

the main process ends when flag_patio_finished turns to True. Now all three child processes are set to daemonic child process by Process.daemon = True, therefore, the child processes will be terminated as soon as the main process completes.

❗️Note that the main process could NOT exit until all queues are closed.

📚 document for multiprocessing.Queue()

📚 Things I Wish They Told Me About Multiprocessing in Python

ANSI codes in webots console

📚 document for AnsiCodes

Keyboard event

📚 document for Keyboard()

Chassis

A four-wheel drived chassis, which mean speed of each wheel is set separately.

Body

Item Measurement Note
Size 0.15m, 0.23m, 0.05m width, length, height
Density $7.85 \times 10^3kg/m^3$ Density of metal

Wheels

Item Measurement Note
Radius 0.033m  
Max Velocity 100rad/s the rover moves forward when velocity is negative
Max Torque 100N⋅m  
Front Track 0.12m Vertical distance from center is 0.068 m
Rear Track 0.12m Vertical distance from center is 0.07 m
Motor Control Method velocity control PID is not used in velocity control. See here for doc

Arm

Item Measurement Note
Density $0.8 \times 10^3 kg/m^3$ Density of plastic
Arm Length 0.035m From axis of Elbow Motor to axis of Wrist Motor
Max Velocity 20rad/s  
Max Torque 10N⋅m  
Holder Size 0.04m, 0.066m, 0.024m width, length, height

Visual & Sensor

📑 Basic usage of several sensors

The illustration of the signals we get: | Name | Data Type | Description | | —————- | ————————— | ———————————————————— | | Position | [float, float, float] | The position of the robot which is same with translation | | Speed | float | A float in m/s | | Bridge_Detection | bool | If the bridge is in the right position, return True | | Gate_Detection | bool | If the bridge is in the right position, return True | | Color | str | The predefined string that indicate the color | | Direction_x | float | Ranges of (-180, 180]. Indicating the degree that the head deviate from x-axis | | Direction_-z | float | Ranges of (-180, 180]. Indicating the degree that the head deviate from -z-axis | | Path_Direction | float/None | Range of approximately [-54.88, 54.88]. A float number that indicates the degree that the path-direction deviates from the head direction. If there is no path, None is returned. |

⚠️ Noticing that, the right deviation is positive and the left deviation is negative. ⚠️ Noticing that, the original signal of bridge and gate detection is False. Once the object is detected, the corresponding signal will turn to True and the detection function will not process again.

Path Camera Parameters

Item Measurement Note
Height 0.553m Height of camera center to the ground
Position 0.1m From center of the rover to the front
depression angle 1.57rad From horizontal axis to down
Field of View 0.9  
Size 138*138 It seems there are some black pixels at the edge, so 5 pixels are cut off at each edge when passed into the program
Actual Size of ROI 0.52m, 0.21m Width and length of the wood board below

Decision

Environment

Specifications

Item Measurement (x, y, z) (m) Note
Patio 100, 2, 30 With wall of height , thickness of 2m, 0.5m
Pond 55, 1.9, 9  
River 100, 2, 2  
Asphalt Road width: 0.2 At height of 2.002. Radius of the Curve is 0.8m
Orange Box 0.1, 1 Width and length. Actual color (r, g, b): 245, 121, 0. height of top is about 8.7mm above the ground, slope of about 7.5 degree
Bridge 0.1, 0.338, 2.8 Slope is 20 degree, made of three 1, 0.1, 1 boxes
Color Box 0.2, 0.6 Width and length
Color Road width: 0.05 Colors: yellow (255, 255, 0), red (255, 0, 0), purple (153, 0, 255). Intersection angle of red and yellow path is 0.25rad
Fish Food radius: 0.005 mass: 0.01g, 18 balls in total. High total mass weaken the steering performance

Development Strategy

A regular automatic system development looks like this👇

But since we don’t have much experience and we don’t have very long time, we do it like this👇, start from the two green circles at the same time, which saves our time and gives us more chances to adjust the design.

Personnel Division

📑 detail

Project Specifications

First a homemade simulation environment, a.k.a. the patio is needed.

The required patio is shown below. Explicit labeled measurements can not be changed.

❗️ the green and red boxes are just for illustration, should not really appear.

Item Measurement
Rover maximum of 50x50 cm
Bridge 100 cm wide, 3 m long
Arch 100 cm wide, 100 cm high (suggested)

Tasks

Task1

From the start point get to the first red box following the line.

💡 although there should not really be a red box, but can be set and measured by distance.

Task2

Release a kiwi into the pond when the orange box is detected.

Task3

Detect the bridge and go across it, the get to the right of the trees.

💡 Here a beacon could be used to avoid tree recognition.

Task4

Detect the arch and get through it. Then follow the line to the color box.

Task5

recognize color of the color box and follow the line in same color to the end.

💡 Color of the color box could be set manually.

报销流程及要求

学院对于本课程采取凭发票报账报销政策，需组员在购买过程中按照学院要求开具增值税发票.

具体报账要求详见 📑报账.md

支出信息公开

本栏目每周更新一次, 旨在进行项目支出信息公开.

具体支出明细详见 📑信息公开.md

This site is open source. Improve this page.