In the increasingly crowded menagerie of robotic quadrupeds, it takes a fairly eccentric gimmick to make anyone look up from their morning brew. HongKong Luwu Intelligent Technology Limited, the brains behind the XGO line of robot dogs, has seemingly decided that the best way to stand out is to embrace a bit of an identity crisis. Their latest offering, the XGO-mini2SW, is a wheeled-legged hybrid that can’t quite decide if it wants to trot or glide, pairing the nimble agility of 12-DoF bionic legs with the frictionless efficiency of wheels. The firm is currently gearing up for a Kickstarter launch.
At the heart of this metallic mutt sits a Raspberry Pi CM5, a proper bit of kit that provides developers with a flexible, open-source playground. This isn’t merely a glorified remote-controlled toy; it’s a serious development platform for embodied AI. The company is pitching the robot as a prime candidate for reinforcement learning, offering a robust open-source “Sim-to-Real” pipeline. In layman’s terms, this allows researchers to train AI models in a digital sandbox before letting them loose on the physical hardware—a vital workflow for anyone serious about modern robotics.
The XGO-mini2SW is also kitted out for multimodal AI interactions, boasting support for voice commands and vision-language processing. For those who want their robot to do more than just scuttle across the lino, there’s an optional 4-DoF robotic arm that can be bolted onto its back. This allows the bot to grasp and move objects, triggered by nothing more than a natural language prompt. Being a staunchly open-source project, the company has tucked all its code away on GitHub for the community to poke and prod: Xgorobot GitHub.
Why should we care?
While the market is currently awash with slick, locked-down proprietary robot dogs, the XGO-mini2SW is aiming for a different crowd: the builders, the tinkerers, and the AI researchers who aren’t afraid to get their hands dirty. By anchoring the tech to a Raspberry Pi and providing open-source tools for high-level techniques like reinforcement learning, XGO is offering an accessible, relatively low-cost sandbox for the future of AI. It’s less of a digital pet and more of a walking (and rolling) laboratory for anyone who believes the best way to understand AI is to give it a body and see how it copes with the real world. Just try not to trip over it when it’s having a mid-afternoon existential crisis.