17. Applications and Implications¶
Here goes a brief summary of the final project developed here.
What will it do?¶
It will take advantage of the features of a robotic arm able to catch objects and artificial intelligence to recognize object of interest. This project aim a self-moving robotic arm capable of recognize the object of interest and catch it automatically. The robotic arm might be used to support people with disability, identifying for example medicine that the patient need to take and bring it to him or her automatically.
Who’s done what beforehand?¶
According to Hackaday.io there are around 44 projects tagged with robot arm in the website. This number does not limit to it when considering the world wide web. In fact, it is possible to find several projects of robotic arm over the internet.
Autonomous robotic arm is a well-know technology being used in diverse segments, for example automobile industry and even medical systems. Most of these projects are programmed to perform the same task constantly to avoid humans to do boring tasks every day. This allows them to focus on other tasks that are not possible to automate improving the productivity of the industry.
The difference between the existing projects and the one being developed here, is the combination of the power of artificial intelligence giving a more generic meaning to a robotic arm. Using the image recognition to identify objects of interest, the robotic arm can be programmed to different tasks. For example, it can be formulate to catch a certain object, for instance a box of medicine, and later drop it in a desired location. The application does not limit to this, it can be used to identify obstacles and explore hostil environment that normally human cannot reach yet.
I would like to mention two projects that I found in my research. MeArm is a well-known robotic arm project used by several purposes. Most of them, it is for educational purposes, but not limited to it. This is an open source robot arm designed using small number of components. It is also cheap to acquire it and to assemble it.
How to Mechatronics is a youtube channel that offers several tutorial of how to design electronic systems. It is possible to find two very nice ideas of how to build a robotic arm build. First, the Static arm that is fixed in the ground and have 4 points of movements. It uses arduino to decide about the movements of the arm and smartphone app to request the arm to move. The second, Mobile arm, it is the same “Static arm” but now integrated in a mobile vehicle. The mobile app has been changed because there was a problem of integrating some components with the arduino.
This project will extract important features from these two applications.
What will you design?¶
I will design an artificial intelligence application that recognize objects read from a camera and also the robotic arm that will be triggered to catch objects of interested. The idea is also to design a vehicle to enable the arm to explore the environment looking for the object. Later, extra functionality can be added, for example the use predefined GPS location of where the object might be located. Or time schedule to catch certain object according to the agenda of important tasks.
What materials and components will be used?¶
I will design the first prototype of the arm in cardboard to make sure the arm has the necessary movements to catch the objects. The second prototype I will design in wood for having more stability and strength. It will be necessary one microcontroller containing ESP8266 (WiFi board), four servos, one raspberry pi, and a camera.
Where will come from? (the material)¶
The lab has some websites that usually buy components in a better price. Here I will list some that you can find them.
How much will they cost?¶
Here it is the average of price of each component:
- 4 servos – each servo 5€, total of 20€
- raspberry pi
- Wemos microcontroller
- distance sensor
- wood: 3€ (50x30 cm and 3mm thickness)
- screws: 3€ (package of screws)
What parts and systems will be made?¶
I will design my own robotic arm inspired by some existing projects, but trying to create a new arm. The arm will be designed in a 2D software (e.g. Inkscape) and later on it will be used a 3D design software but, I will use it make the pieces using a laser cut machine. However, some parts of the arm will need to be designed in 3D printer, for example a support for the raspberry pi camera. It will also be necessary a PCB to connect the servos to the microcontroller.
I will also create a system that will be running the image recognition from a raspberry pi camera. There will be an mobile application where the user might use to select the type of object he or she wants to catch with the arm. The modules of the project will communicate with each other via MQTT using the WiFi network.
What processes will be used?¶
It will use tensorflow for image recognition. It will use MQTT to enable communication between the modules of the project. It will use Wemos to control the robotic arm. It will use AppInventor to create the mobile app that will configurate the robotic arm.
What questions need to be answered?¶
- Can a robotic arm be configurated to have different meanings?
- Can a robotic arm take advantage of AI for identifying object of interest?
- Can a generic robotic arm work autonomously?
How will it be evaluated?¶
- The arm should have correct moviments to enable catching the object.
- The artificial intelligence process should trigger the robotic arm to catch the object.
- The arm should recognize the distance from the object to be able the move to catch it.