In a video just released today, the Minitaur robot shows how to deal with various terrains by dynamically adjusting the gait. It can crawl, jump, walk on the ice, walk upright with two legs, and many other new skills!
Ghost Robotics co-founders Gavin Kenneally and Avik De told us that one of their main goals is to expand Minitaur's behavior so that robots can “pass through a wide range of terrain environments and real-world operating scenarios,†he added, and Ghost Robotics also "The legged robot is considered not only to have superior baseline mobility in various environments and terrain relative to wheeled robots and orbital robots, but also to exhibit a series of different behaviors that make it easy to overcome natural obstacles."
For more details, Lei Feng has selected the contents of the in-depth conversation between IEEE and Kenneally and De and Ghost Robotics CEO Jiren Parikh for your reference.
IEEE Spectrum: What is the difference between a legged robot like Minitaur and a wheeled or orbiting robot?
Ghost Robotics: On a flat surface with no obstacle path, wheeled robots are more efficient than tracked, legged, and even aerial robots. In sand, mud and rough terrain, tracking robots outperform wheeled robots. However, with fixed objects, obstacles and vertical surfaces and no alternative paths, the legged robots are unparalleled in comparison to the tracking robots. Even if moderate objects and obstacles can be overcome by the tracking robot, the energy efficiency of the tracking robot on a large stretch of unstructured terrain will be reduced compared to a dynamic legged robot.
Another advantage is that legged robots typically have more drive freedom than similarly sized tracking or aerial robots (and those additional drive degrees of freedom) can be applied to things like reorientation, manipulation and making the robot more flexible Ways and more general ways to get around. In scenes such as beach and mudflat, the tracking device can do well before reaching the trapped point, but their escape ability is very limited, and the Minitaur will have greater mobility and ability to escape. In addition, in order to gain the ability to open the door, you must attach a robotic arm to the body of the wheeled, tracked or aerial robot, and we have shown that Minitaur has this ability without any physical modification.
The main challenge with legged robots is multi-degree of freedom coordination and balancing on various terrains. Minitaur is specifically designed to allow very flexible and versatile software control of its limbs at high frequency bands, which allows control designers to advance free and continuously improved control algorithms over time (without the need to modify the robot body).
"Direct drive actuators are inherently strong because there is no gear device that breaks due to impact loads and there are no hydraulic systems or force/torque sensors that could be damaged."
IEEE Spectrum: What is your experience with Minitaur's durability when doing outdoor dynamic tests?
Ghost Robotics: Considering that Minitaur is still in trial production, we have been conducting detailed physics experiments on robot prototypes since day one. This is not easy (and full of fun). And it's not easy to repair on site because of machine damage caused by some methods. Dropping from a height may bend the aluminum legs, but these can be easily corrected or replaced without losing any functionality. The casing is quite rugged even in the current design state, and the motor is fully protected, so the mechanical leg becomes a primary concern. The direct drive actuator itself is very robust because no gear device is broken due to impact loads and there is no hydraulic system or force/torque sensor that can be damaged.
One of Minitaur's core design principles compared to other legged and tracked robots is its low mechanical complexity. The tracking robot looks simple, but the suspension mechanism is very complicated, which makes it difficult to repair after damage. If the tracking robot's tread or suspension mechanism is damaged during travel, it can only rotate in place, but if the Minitaur's leg is bent or fixed, it can continue to stagger forward.
IEEE Spectrum: Can you describe how the Minitaur changes gait to suit different types of terrain?
Ghost Robotics: A very basic example is that gait is designed to respond to disturbances with real-time feedback (such as the toes slipping on ice, or the unevenness of walking on a rock bed). If you look closely at the Minitaur's video walking on the ice, you can see that when the robotic leg starts to slip, it will circulate faster and always swing and reposition under the body to prevent the body from falling to the ground. The conventional design of multi-leg walking is to use a fixed "clock" signal to trigger a fixed frequency (and usually along a fixed trajectory) to move the leg. Obviously, keeping the robotic legs under the body without feedback from the legs and the environment when the robotic legs begin to slide and move with a rigid motion architecture would be very challenging, if not impossible.
IEEE Spectrum: Your video shows that Minitaur uses different creative ways to cross different terrains. What kind of multimodal sports are you working on?
Ghost Robotics: We've already shown in our first video a fence climbing with a mechanical toe attachment and intend to demonstrate in the upcoming video the use of fixed leg attachments to climb a variety of vertical surfaces. According to different application examples, we expect that the accessories of the mechanical legs can be interchanged in the field in the future design.
We are also customizing with a private client to use Minitaur as a surface and underwater swimming robot or submersible platform to work with ankles on the bottom of the seabed or bed. If you look at the process of the Minitaur's mechanical leg retraction, you will understand how we can accomplish water seal and stability and air bubble control buoyancy design without investing too much. Our robot is relatively quiet (no gearbox working noise), which makes it widely used in scientific research and military applications, and also has a very high specific power (this is one of the limiting resources of underwater vehicle installations). .
"If the tracking robot's tread or suspension mechanism is damaged during travel, it can only rotate in place, but if the Minitaur's leg is bent or fixed, it can continue to move forward."
IEEE Spectrum: How far can Minitaur use only two legs? Is it possible to control with the other two legs while maintaining balance?
Ghost Robotics: The study of biped walking is an ongoing and very challenging job - we don't think there is another 3D bipedal robot in the world with only 4 drives. Of course, we are very satisfied with the progress we have made. The Minitaur can walk upright for nearly 20 steps (using two legs), and then when it knows that it can no longer maintain a balanced state of the two feet, it can be changed back to four feet walking. We will continue to study and plan to keep it as long as possible.
Using one or two legs for various tasks is the key behavior of the Minitaur we are studying (for example, opening the door). Biped use cases include object manipulation, robot positioning to climb vertical surfaces, gaining favorable positions to read sensor data, escape exercises, and supporting/linking to assist in balance.
IEEE Spectrum: What is the outlook for extending Minitaur to a mid-sized (or larger) legged robot?
Ghost Robotics: We have pushed the basic limit torque density of commercial motors currently in use. The selected motor is crucial for keeping the robot at a low price, even lower than existing tracked robots, and at a much lower cost than other legged robots. Using current engine technology, we cannot create a 40-cm-long direct-drive robot that remains the same as the Minitaur robot. We can build a more cumbersome version of the same range and have a better payload capacity of the robot, but we must make sacrifices to increase the length. When cost is not a major issue, we are also considering modifying/customizing the motor design for future specific use cases.
Despite this, our ongoing design work allows us to downsize Minitaur, providing the same functionality with a smaller case and payload capacity when a specific application requires a smaller-sized robot. Please continue to pay attention to this aspect.
Don't worry, Ghost Robotics, we will wait and see.
Oh, just a kind reminder, Minitaur is true, it's really cheap, only about $ 10,000 a, this is the price of the hand instead of mass production. Ghost Robotics is trying to convert Minitaur from a developer-oriented development platform to a commercial application version, which also includes support for more sensors, operational capabilities, greater robustness, and some built-in automation scenarios.
In fact, as Xiaobian knows, they are busy reducing the size of the robot while also actively reducing costs, which makes me look forward to the future - Ghost Robotics may launch the Mini Minitaur so that robot lovers like me can easily have . If I haven't expressed it very clearly: I love this thing and desperately want to own one.
Germanium single crystal has high hardness, good thermal conductivity, insoluble in water, and the transmission band is 2-12 μ m. It is a very common infrared optical material. At present, it is widely used in infrared imaging system and infrared spectrometer system. Germanium single crystals are also used as substrate materials for various infrared filters.
Germanium Window
Hanzhong Hengpu Photoelectric Technology Co.,Ltd , https://www.hplenses.com