[ad_1]
Quadruped Robots: Already Here and Getting the Job Done
The use of robots has become an increasingly popular trend in the industrial sector with bipedal humanoids and quadrupeds being at the forefront. Quadrupeds, in particular, have shown potential for extensive use in multiple fields such as inspections in power plants and refineries, playing soccer, and even providing security services. While the development of bipeds is still underway, quadrupeds are already on the job.
Notably, Boston Dynamics’ Spot has become the poster child of quadrupeds, and several start-ups and research institutions have also come up with their own take on this category, with even smartphone manufacturer, Xiaomi, in on the act. As several quadruped versions continue to develop, Google’s DeepMind recently published a research paper that aims to establish a benchmarking system to quantify the performance of these machines.
Barkour: A Benchmarking System for Quadruped Robots
Google’s DeepMind research paper on Barkour suggests that, for a long time, there has been no baseline benchmark to determine the efficacy of quadruped robots. Given that quadruped robots are inspired by animals, the research team believes that using real animals would provide the best performance analog for their robotic counterparts. To achieve this, the team set up an obstacle course in the lab, with a dog running it. It turned out the little wiener-dog was as tenacious as they come.
The Barkour Obstacle Course
The Barkour obstacle course is composed of four obstacles, densely packed within a 5×5-meter area -much denser than traditional dog shows. The course is designed to test the robot’s agility, speed, and functionality in crawling, running, balancing, and jumping. The performance of the quadruped robots is rated on a scale of 0 to 1, with a simple binary, determining whether the robot can cross the space in approximately ten seconds, like an average-sized dog. Penalties are imposed for slow speeds, skipping obstacles, or failing to complete obstacles.
Barkour: A Challenging Benchmark
The results have been impressive, with Google stating that Barkour has become a challenging benchmark that can be easily customized. Furthermore, the learning-based method for solving the benchmark provides a quadruped robot with a single low-level policy that can perform a variety of agile low-level skills. The research team believes that developing a benchmark for legged robotics is vital in quantifying progress toward animal-level agility.
Conclusion
Quadruped robots have shown their potential in various fields, including offering inspection services in hazardous industrial environments and sports such as soccer. However, the lack of a benchmarking system that could quantify the efficiency and agility of the robots has been a challenge. The Barkour obstacle course, developed by the researchers at Google’s DeepMind, provides an effective way to measure how well these machines perform. The course acts as an analog performance benchmark for quadruped robots and has proved to be a crucial step forward in developing robots capable of animals’ level of agility.
FAQ
What is a quadruped robot?
A quadruped robot is a robot with four limbs that enable it to walk like an animal. These types of robots show great potential and are being developed for various purposes, including conducting inspections in hazardous industrial environments and entertainment purposes.
What is the Barkour obstacle course?
Barkour is a benchmarking system developed by Google’s DeepMind, which proposes a performance analog for quadruped robots, with real dogs going through an obstacle course, which the robot must replicate.
What tests were in the Barkour obstacle course?
The Barkour obstacle course was made up of four obstacles densely packed within a 5×5 meter area and designed to test a robot’s agility, speed, and functionality. The course featured hurdles for jumping, narrow spaces where a robot could pass through, and areas for balancing.
[ad_2]
For more information, please refer this link