Researchers develop LoCoQuad, an arachnoid-inspired robot

9 Apr 2020

The four-legged arachnoid-inspired quadruped robot is capable of walking, turning, swinging, waving and even take pictures. It can be useful in research and education in the future. (Image Credit: Bernal * Civera, YouTube)

Two researchers, Manuel Bernal-Lecina and Javier Civera Sancho, from the University of Zaragoza in Spain, have developed an arachnoid-inspired quadruped robot named LoCoQuad, which could be used in numerous applications for robotics research and education. The team presented their findings in a pre-published paper on arXib. The open-source robot has incredibly low hardware costs, ranging from $150 to $165 USD, and its software and design can be found on GitHub.

After conducting research on inexpensive robots and creating several 3-D printed components, Bernal-Lecina began a collaborative effort with Javier Civera to develop a new robot. Their goal was to create an inexpensive robot platform that could serve as a benchmark to train and evaluate state-of-the-art, reinforcement learning algorithms.

LoCoQuad is a robotic platform that can walk, turn, wave and even swing by actuating over its four legs. The robot has also been equipped with several sensors, a frontal camera, an inertial measurement unit (IMU) with an accelerometer, gyroscope and two ultrasonic depth sensors. They also use Raspberry Pi as a control unit, which allows the sensors to be flexible and extended.

All the sensors and electronics are placed in the main body of the robot. The battery pack is placed on the bottom, power and driver boards are in the center, and the control board and sensors are on the top. This design enables the robot to achieve balance. All four of the robot legs are connected using MG90S servo horns inserted on custom 3D staples. In order to connect each link, the team designed different 3D parts to fit various ranges of motion and a wide range of positions and configurations. In its basic design with two joints per leg (rotator and knee), LoCoQuad contains four different configurations per knee. There are 16 configurations (4 at the elevator and 4 at the knees) in another design, which consists of three joints per leg (rotator, elevator and knee). These different mechanical variations allow the robot to change and adapt to different shapes.

The four-legged arachnoid-inspired robotic platform’s low hardware costs could allow it to be easily manufactured on a large scale. Even though it’s not the most efficient quadruped robot to date, the fact that it’s inexpensive, open-source, highly configurable and user friendly, makes it a promising platform for research and education purposes.

Both researchers demonstrated the LoCoQuad’s effectiveness in a series of experiments. By doing so, they discovered that it could carry out a wide range of basic tasks, which suggests that the platform could be used in an effort to test out reinforcement learning and different machine-learning algorithms. One of which includes reinforcement learning models, which are being used to develop more efficient and smart robots. LoCoQuad could be used to test these techniques and identify ways for it to be improved.

The researchers have already started to use the quadruped robot in other studies to compare different algorithms and identify which ones are most effective. They also hope that LoCoQuad’s design and software will be downloaded by other teams around the world to test out their own machine learning algorithms.