Tesla is ramping up hiring for Optimus development. At the moment, the company’s goal is not just to create a mindless machine, but a bipedal robot that will bring real daily benefits, taking into account the needs of the owner.
Tesla has further increased hiring for the team developing its bipedal robot, Optimus. What is remarkable is that the company is increasingly tapping into engineers who can make the robot useful. Previously, engineers specializing in robotics and the physical creation of the robot and its mechanisms were more required. In recent months, we have seen significant progress in this, as Optimus can walk, navigate in space, accurately perform delicate work with the help of in-house developed actuators, and even do yoga. However, now more and more vacancies have begun to appear related to making the machine useful, teaching it to perform a wide range of tasks, as has been noticed by Electrek.
Tesla is trying to train bots for various tasks. It uses the same neural network-based strategy used for Full Self-Driving (FSD). In the summer, Elon Musk demonstrated how his car with the installed FSD V12 drove through the streets. He explained that the AI team does not write code to make the car do what it needs to do. Instead, they train neural networks (NNs) by showing them videos of how to drive correctly in various situations. NNs learn and then apply the knowledge while driving.
Among the vacancies for which recruitment is open are a Simulation Software Engineer, Reinforcement Learning Engineer, SLAM Software Engineer, and about ten other positions related to the Optimus program.
Simulation Software Engineer will be in “a unique position to accelerate the pace at which the Tesla Bot Autonomy stack improves over time.” The main ways in which the simulation team realizes this include:
- Building core components to virtualize all aspects of Tesla Bot into its Simulator.
- Testing all Tesla Bot software releases for regressive behavior.
- Generating synthetic data sets for neural network training.
Simulation Software Engineer will contribute to the development of the simulation by building models and simulation tools that are capable of virtually prototyping the system for fast, iterative development and robust validation, the company said.
The goal of the Tesla reinforcement learning team is “to build and demonstrate a general robot learning system that can leverage AI to perform complex physical tasks, ranging from full body locomotion, precise manipulation, and more.” Reinforcement and imitation learning engineers of the company will be responsible for end-to-end robotic learning and own this stack from inception to deployment.
Tesla indicated that SLAM Software Engineer “will marry cutting-edge deep learning algorithms with robust real-time software. You’ll deliver safety-critical features to hundreds of thousands of vehicle customers, or to Humanoid robots operating various useful tasks within our factories.”