Ricky Ye is the CEO of DFRobot, a provider of robotics and open-source hardware, dedicated to creating innovative and friendly products.

While all online tutorials can teach ordinary people how to build robots, we hope that engineers with a certain development background can build a robot from scratch, spending less money.

Let’s take a look at what it really takes to create a robot that meets business needs and has clear functionality.

▍Step 1: Set Your Idea

The first step is to set an idea for the robot. Will it autonomously create beach art? Will it be a family companion with advanced emotional intelligence that learns new things? Cultivating interest in STEM (science, technology, engineering, mathematics) education? Simplifying the manufacturing supply chain process? Sometimes the most successful robots are those that serve a single purpose. In any case, without a clear intent, it is impossible to build a successful creation.

▍Step 2: Choose Your Platform

Next, choose what operating system your robot will run. The best operating system depends on whether you are building an industrial robot or a cognitive robot. Windows 10 surpasses Linux for robots used for daily tasks, mainly due to its integration with Microsoft’s AI solutions. Additionally, Windows 10 offers a better user experience and is easier to operate for those with basic computer knowledge. Once you decide what kind of robot to create, you can choose the operating system that best matches it.

▍Step 3: Build the Brain

The next step, arguably the most important, is to establish the command center of your robot. Whether it’s voice recognition, facial recognition, motion detection, or other features, these can serve as the backbone to support the functionalities required by the robot.

This “brain” can be as simple as handheld internet-connected hardware. Popular iterations include the affordable, programmable microcomputer Raspberry Pi, and the first manufacturer board LattePanda, which comes with a pre-installed full Windows 10 operating system, providing the same user experience as a regular Windows PC. The LattePanda is palm-sized, Intel-based, supports Arduino, and allows for processor optimization and coding for developing custom robot prototypes. The good news is that LattePanda is not only suitable for software developers but also allows ordinary employees to learn step-by-step through tutorials.

▍Step 4: Create the Shell

In Step 1, the functions determined for the robot will dictate its required physical features. For example, if you are building a cleaning robot that requires vision, the brain will need to be compatible with laser scanners, motor drivers, and touch sensors. If the robot needs to communicate with other devices, ensure the brain can adapt to messaging connections.

▍Robots Are Going Mainstream

Some of the most impressive robots have been invented. For example, InMoov—the first open-source, 3D-printed, walking, and communicating humanoid robot. InMoov comes in various shapes and sizes typical of humanoid robots in movies, providing different services for businesses and consumers.

Robots are everywhere; they are not going away. Businesses in North America ordered nearly 35,000 robots last year, with observable growth in productivity and a 10% drop since 2015 in robot production speed. Similarly, by 2020, more than one in ten American households is expected to own a robot. These statistics highlight the need for more understanding of the construction of all mainstream robots, whether basic or complex.

To develop a business-based robot, it is necessary to formulate a clear strategy, weigh the pros and cons of different operating systems, program the brain to perform the desired functions, and build the physical features required to achieve the robot’s functionalities.

References: VB

Editor: Leite

★ Recommended Reading ★

Li Feifei Selected as 2017 ELLE Women in Technology Star

Director of Baidu Silicon Valley AI Lab Explains: Which Hardware is More Valuable in Deep Learning

Andrew Ng’s New Achievement: Machine Learning Models Can Diagnose Arrhythmia More Accurately Than Doctors

Collection ‖ Comprehensive Machine Learning Quick Reference Guide

Download ‖ 2017 CVPR Open Paper Collection