Unitree's new A2 quadruped gets dual lidars and serious climbing skills

Just a week after shocking the robotics world with a $5,900 humanoid named R1, Unitree is back to upgrading its four-legged lineup.

The Hangzhou-based robotics company just launched the Unitree A2, its latest industrial-grade quadruped. Unitree, which raised a Series C round last month that pegged its valuation at around $1.7 billion, designed the A2 specifically for real-world deployments across inspection, logistics, and research and development.

Eyes in the back of its head

The most noticeable upgrade on the A2 is how it sees the world.

Unitree packed the robot with two industrial lidar sensors, mounted at the front and rear, paired with an HD camera and a front headlight. The previous B2 quadruped only had a single lidar alongside depth and optical cameras. By doubling down on lidars, the A2 delivers 360-degree 3D depth perception and eliminates nasty blind spots.

The hardware is tied together by an AI vision system capable of depth sensing and autonomous obstacle avoidance, designed to help it navigate messy worksites without human intervention.

Lighter body, bigger steps

Instead of making the A2 heavier and more powerful than the B2, Unitree chose a smarter tradeoff: shedding weight to conquer rougher terrain.

The A2 weighs in at 37 kg unloaded, a dramatic drop from the hefty 60 kg B2. While the B2 still beats it on raw top speed (6 m/s vs. 5 m/s) and maximum standing load (120 kg vs. 100 kg), the A2 shines in obstacle clearing. It can step over obstacles up to 1 meter high, more than double the 40 cm limit on the B2.

When it comes to hauling gear, the A2 handles a walking payload of 25 kg. Fully loaded, it walks for 3 hours or roughly 12.5 km. Without cargo, its range extends to 20 km. It also features hot-swappable dual smart batteries, meaning field operators can swap power packs on the fly without shutting down missions.

Naturally, any quadruped faces comparisons to Boston Dynamics' Spot. On paper, the A2 outmuscles Spot, which maxes out at a 14 kg payload and a 1.6 m/s top speed. However, Spot still holds a clear edge when it comes to mature, built-in autonomous software.

Why it matters

We think Unitree made a very clever pivot here.

In industrial inspection, raw sprint speed is useless if your robot gets stuck on a pipe or tumbles down an unmapped industrial staircase. By cutting the weight down to 37 kg and boosting step climbing to 1 full meter, Unitree addressed the two biggest headaches for site managers: mobility over awkward obstacles and human handleability when things go wrong.

The hot-swappable battery setup is another clear signal that this is built for round-the-clock shift work, not just tech demos.

That said, winning on hardware specs is only half the battle. While Unitree can out-spec Boston Dynamics on payload and speed at a fraction of the cost, Spot's true moat is its battle-tested autonomy and rock-solid navigation software. Unitree eliminated physical blind spots with front and rear lidar, but it still has to prove its AI software stack can match the reliability of Western alternatives in high-stakes environments.