Remember MIT’s robotic cheetah from years ago? Well, it finally got upgraded, and now the robot can run faster than ever. Researchers from MIT’s Improbable AI Lab, part of the Computer Science and Artificial Intelligence Laboratory (CSAIL) and directed by MIT Assistant Professor Pulkit Agrawal have spent their time working on fast-paced strides for the robotic animal. It’s safe to say that it was a success, as their model-free reinforcement learning system broke the record for the fastest run recorded.
To achieve this, the researchers had to find a way to ensure that the robot adapts to its environment, including quickly identifying terrain changes and adapting to prevent itself from falling over. This means that the robotic cheetah has to learn by trial and error, thus removing the need for a human to specify precisely how it should behave in every situation. What they ended up with was an upgraded robot that hit 8.7 mph.
- Space vehicle for boys and girls – This space toy for exploration features steering, a raising and lowering function, a crane and an operator’s...
- Raise and lower the crane – Kids can raise the operator’s cab and raise and lower the crane to pick up and move the cargo
- Adapt to different terrains – Young space explorers can put the loader to the test as they imagine what it would be like to live and work on another...
Programming how a robot should act in every possible situation is simply very hard. The process is tedious, because if a robot were to fail on a particular terrain, a human engineer would need to identify the cause of failure and manually adapt the robot controller, and this process can require substantial human time,” said MIT PhD student Gabriel Margolis and IAIFI postdoc Ge Yang.
