PhysEvo: Astra Can Act, Let It

Wenqing Tian, Zeyu Zhang, Zhaocheng Liu, Fengwei Liu, Qiang Liu, Liang Wang · N/A · 2026

Summary

Abstract Summary

Astra can act, yet reliable manipulation depends on the system through which it observes and controls the world. We introduce PhysEvo, a framework for physical recursive self-improvement (RSI) around a single frozen model. A task agent executes robot tasks; a meta-agent uses the resulting trajectories to diagnose failures, revise tools and skills, and test corrections. The meta-agent can also improve its own diagnostic tools, so retained revisions support both later action and later self-improvement. This process develops joint-level control, evidence-seeking observation, and reusable manipulation skills without model-weight updates or a separately trained action policy. Across 42 RoboDojo tasks, held-out-layout evaluation of retained task-specific deployment versions yields a five-dimension average score of 68.14/100 and 62.00% success, compared with 47.17% for RoboDawn's one-shot Astra agent, the strongest published reference in our comparison. On eight manipulation tasks challenging direct Astra, PhysEvo achieves 55.00% success, compared with 1.25% for the direct-Astra reference. Deploying the simulation-evolved harness on AgileX PiPER and continuing skill revision yields 90.60/100 average score and 84.00% success across 25 trials on five real-world tasks. PhysEvo turns the consequences of action into persistent, testable changes to how a frozen model acts and improves.

Key Points

Astra can act, yet reliable manipulation depends on the system through which it observes and controls the world.
We introduce PhysEvo, a framework for physical recursive self-improvement (RSI) around a single frozen model.
A task agent executes robot tasks; a meta-agent uses the resulting trajectories to diagnose failures, revise tools and skills, and test corrections.
The meta-agent can also improve its own diagnostic tools, so retained revisions support both later action and later self-improvement.
This process develops joint-level control, evidence-seeking observation, and reusable manipulation skills without model-weight updates or a separately trained action policy.

PhysEvo: Physical Recursive Self-Improvement

PhysEvo alternates between acting in the environment and revising the system used to act. A single frozen model performs both roles. Physical outcomes provide evidence; tested changes to the harness preserve what is learned from that evidence ( figure 2 ). Let $M_{\theta}$ denote Astra with fixed parameters.

At iteration $r$, the persistent system consists of a task harness $H^{\mathrm{task}}_{r}$ and a meta harness $H^{\mathrm{meta}}_{r}$: The task harness supplies observations, action tools, execution feedback, and skills.

PhysEvo: Physical Recursive Self-Improvement — PhysEvo: Astra Can Act, Let It — Figure 2: Physical RSI in PhysEvo. A frozen Astra acts through the task harness and revises both task and meta resources from execution feedback. Accepted revisions support later episodes and improvement rounds.

The Evolved Embodied Framework

Physical RSI reshapes what Astra can control, what evidence it can obtain, and what guidance persists. We describe the resulting architecture below; tasks retain their own deployment versions ( section 5.1 ). The default control_arms interface accepts twelve absolute arm-joint targets and two continuous gripper commands. It supports both arms and remains available while an object is held.

The executor enforces native actuator limits and checks measured motion; Astra chooses the approach, arm configuration, and reconfiguration.

The Evolved Embodied Framework — PhysEvo: Astra Can Act, Let It — Figure 3: Joint reconfiguration enables recovery from an obstructive arm posture.

Experiments

We first evaluate retained deployment versions on held-out simulation layouts, then examine the revisions behind the resulting framework. On real hardware, we continue skill adaptation after deployment. We evaluate all 42 base RoboDojo simulation tasks across generalization, precision, long-horizon execution, memory, and open-ended instruction following.

Native score ranges from 0 to 100; success rate (SR) counts scores of 100. Each task averages one attempt on each of five held-out layouts.

Experiments — PhysEvo: Astra Can Act, Let It — Figure 4: Skill revision maintains a useful view throughout pouring and recovery.

Sources and Demonstrations

Figures are reproduced from Tian et al. See the full paper for experimental details and the project page, when the paper links one, for demonstrations and videos.