AIVOT builds intelligent robotic systems to perform various complex tasks in dynamic environments using the latest AI advancements. The robots can navigate autonomously with their mobile bases and perform dexterous tasks using their two arms. They learn new skills by listening to voice commands and watching human demos. The robots communicate with operators using spoken natural language. The robots perceive the dynamic world state using vision, detect objects of interest, understand the task context, and plan actions in real time to accomplish the specified high-level goals. AIVOT’s end-to-end software stack can be deployed on any off-the-shelf commercial hardware of any size and type, such as robotic arms, quadruped robots, and unmanned vehicles. The robots do all the processing onboard without requiring a network connection.