Repository navigation
We are releasing v0.1, the first public release of RPent.
RPent (Recursive Physical Agent) is an open-source infrastructure for embodied agents in the physical world. It connects general-purpose models, VLA models, memory, and robot interfaces to carry out physical tasks through planning, execution feedback, and experience reuse.
v0.1 supports agent harnesses such as Codex and Claude Code for task planning. It integrates the LIBERO family, RoboCasa365, RoboTwin, and RoboDojo simulators, as well as single-arm Franka, dual-arm Franka, and YAM robots, and provides Memory, Flash Mode, an interactive CLI, and a Web Dashboard. RPent has been evaluated in simulation and on real robots, with leading task success rates on benchmarks such as LIBERO-PRO among the methods listed on the leaderboard.
Highlights
Real-World Demos
- Robot task demonstrations: Dual Franka adapts tableware placement to task goals and performs steel-ball pouring, plate wiping, and storage tasks. YAM sorts drinks by brand labels, checks grasp status, and moves occluding objects to find a spoon. Dual Franka demos | YAM demos
Simulation Benchmarks
- Multi-environment evaluation and a public leaderboard: RPent with GPT-6 Astra achieves success rates of 92.63% on LIBERO-PRO (low reasoning) and 59.20% on RoboCasa365 Target50 (high reasoning). RPent with GPT-5.5 achieves 62.4% on RoboTwin C2R. The leaderboard includes evaluation configurations, success rates, runtime, and token usage. Leaderboard
System Design
-
Modular architecture: An open, modular design separates the system into user, intelligence, interface, and environment layers.
-
Unified interfaces: VLA models and analytical primitives are exposed as tools through MCP or native tool calling. RPC connects model, robot, and simulator services. Core interfaces
Memory
-
Accumulating and reusing experience: Exploration accumulates experience. Evaluation uses read-only memory organized into general knowledge, task strategies, and execution records. Memory
-
Faster execution with task cards: Flash Mode turns successful workflows into task cards and updates action targets using current visual observations, reducing step-by-step planning overhead. Across 800 LIBERO-PRO episodes, it achieves a 72.63% success rate with an average runtime of 60.19 seconds, approximately 6.8× faster than Codex + GPT-6 Astra (low reasoning). It uses the officially released GPT-5.5 exploration memory; the reported scores are the best evaluation results from seeds 0–9. Flash Mode | Evaluation data
Research
- Harness VLA: RPent’s first research paper describes how task memory combines VLA models with analytical primitives to handle task changes, environmental perturbations, and multi-step manipulation without updating VLA weights. Paper | Project page
Features
Real-World Robots
-
Single-Arm and Dual-Arm Franka: Arm and gripper control, multi-camera observations, and hand-eye calibration. Single-arm Franka can connect to an external VLA grasping service. Dual Franka provides grasping, handover, and placement skills, with two-node deployment and operator-assisted exploration. Both support replaying human-reviewed Flash task cards, with operators confirming the outcomes. Single-arm | Dual-arm
-
YAM: Bimanual end-effector motion, wrist rotation, gripper control, multi-camera observations, and Pi0.5 action models. Exploration, human feedback, and experience storage are available through the terminal or Dashboard. Guide
Simulators
-
LIBERO and LIBERO-PRO: Spatial, Object, Goal, and LIBERO-10 task suites, with instruction and scene-layout perturbations, exploration, memory-based evaluation, and Flash Mode. Guide
-
RoboCasa365: Navigation, mobile-base motion, and arm manipulation with PandaOmron, including Target50 evaluation configurations. Guide
-
RoboTwin: Bimanual simulation with
demo_cleananddemo_randomizedscenes, multi-camera observations, and evaluation configurations for 50 tasks. Guide -
RoboDojo: An Isaac Sim / IsaacLab simulator with the ARX-X5 dual-arm platform. Installation is currently supported. Guide
Models
-
Planners and models: Codex, Claude Code, and RPent’s built-in planner can connect to models such as Claude, GPT, and Qwen, with planner-specific reasoning effort, turn limits, and timeouts. In LIBERO, RoboCasa365, and RoboTwin, planners can also output action control values for the environment to execute directly. Configuration guide
-
VLA models: Pi0.5 for LIBERO, RoboDojo, Dual Franka, and YAM; RLDX-1 for RoboCasa365; and LingBot-VLA for RoboTwin. Pi0.5 | RLDX-1 | LingBot-VLA
-
Perception models: SAM3 for object segmentation and Molmo2-8B for visual grounding, combined with depth and camera information to locate targets. Visual grounding guide
System
-
Extension interfaces: Integrate new robots, models, and tools while reusing service communication, task execution, and memory management. Developer guide
-
Interactive CLI and Dashboard: Task interaction and saved run records. The Dashboard displays planning, tool calls, camera views, and action timelines, with English and Chinese interfaces. CLI | Dashboard
-
Service deployment: Model and environment services can run independently on local or remote machines. Consecutive tasks can reuse running model services. Deployment guide
-
Memory: Public datasets of task experience and task cards, with local memory validation, merging, and index updates, allow agents to reuse existing experience. User guide | Dataset
RPent Benchmark Results
See the evaluation documentation for public benchmark results and configurations. Release validation results are included at the end of this page.
Getting Started
-
Run your first task: Start with LIBERO-PRO, install the environment, download Pi0.5 and SAM3 weights, and configure a planner. Installation and quick start
-
Explore and reuse experience: Generate Memory through exploration or reusable Flash execution plans from successful trajectories. LIBERO exploration | Flash Mode
-
Deploy on real robots: Follow the platform guides to install robot control software, configure cameras, complete hand-eye calibration, and obtain and configure the required model weights. Franka | Dual Franka | YAM
RPent v0.1 Test Results
RPent v0.1 was validated with 180 release-version episodes, 180 sampled paper-reproduction episodes, and 15 memory-reuse episodes across LIBERO-PRO, RoboCasa365, and RoboTwin, plus three real-world tasks on Dual Franka and YAM.
One execution of a task with a specified seed counts as one episode. Simulation tasks use each environment’s native success criterion. Failures and timeouts count toward the total episode count. Success rates are the proportion of successful episodes. Runtime and token usage for release-version and paper-reproduction tests are per-episode averages; aggregates are weighted by episode count. These results cover the selected release-validation tasks. See the Leaderboard for full benchmark results.
1. Release-Version Simulation Tests
Tests use RPent release/v0.1 and HF Memory release/v0.1, with read-only memory during evaluation.
1.1 LIBERO-PRO
Codex + GPT-6 Astra (low) with Pi0.5. Eight Task and Swap suites cover Spatial, Object, Goal, and LIBERO-10. Each suite includes two selected tasks, each run with seeds 1 through 5, for 80 episodes.
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| libero_spatial_task · task 0 | 5 | 5 | 100% | 288.54 | 1,469,403.20 | 4,355.00 |
| libero_spatial_task · task 4 | 4 | 5 | 80% | 348.88 | 1,587,509.80 | 4,888.00 |
| libero_spatial_swap · task 0 | 5 | 5 | 100% | 212.86 | 741,139.60 | 3,241.20 |
| libero_spatial_swap · task 4 | 3 | 5 | 60% | 339.52 | 1,370,942.40 | 4,389.80 |
| libero_object_task · task 0 | 5 | 5 | 100% | 180.36 | 726,203.80 | 2,863.00 |
| libero_object_task · task 4 | 5 | 5 | 100% | 192.32 | 929,688.80 | 3,153.00 |
| libero_object_swap · task 0 | 5 | 5 | 100% | 251.54 | 1,311,042.20 | 3,873.80 |
| libero_object_swap · task 4 | 5 | 5 | 100% | 247.34 | 1,057,305.80 | 3,253.80 |
| libero_goal_task · task 0 | 3 | 5 | 60% | 410.84 | 1,657,833.40 | 4,453.60 |
| libero_goal_task · task 7 | 5 | 5 | 100% | 143.88 | 434,249.00 | 2,133.20 |
| libero_goal_swap · task 0 | 5 | 5 | 100% | 523.70 | 1,462,251.80 | 3,985.20 |
| libero_goal_swap · task 7 | 5 | 5 | 100% | 213.40 | 316,889.40 | 1,931.20 |
| libero_10_task · task 3 | 5 | 5 | 100% | 507.38 | 2,201,277.20 | 5,133.20 |
| libero_10_task · task 9 | 3 | 5 | 60% | 469.48 | 2,895,074.00 | 6,348.40 |
| libero_10_swap · task 3 | 5 | 5 | 100% | 518.64 | 3,341,025.40 | 6,639.00 |
| libero_10_swap · task 9 | 1 | 5 | 20% | 715.74 | 5,376,288.20 | 9,809.60 |
| Total | 69 | 80 | 86.25% | 347.78 | 1,679,882.75 | 4,403.19 |
1.2 RoboCasa365
Codex + GPT-6 Astra (low) with RLDX-1. Three Atomic tasks use seeds 1 through 10, and four Composite tasks use seeds 1 through 5, for 50 episodes.
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| Atomic · OpenDrawer | 5 | 10 | 50% | 261.97 | 504,529.00 | 2,066.10 |
| Atomic · NavigateKitchen | 10 | 10 | 100% | 86.55 | 154,368.10 | 854.40 |
| Atomic · PickPlaceCounterToCabinet | 8 | 10 | 80% | 186.62 | 318,589.00 | 1,537.20 |
| Composite-Seen · PrepareCoffee | 0 | 5 | 0% | 697.10 | 1,318,701.80 | 4,115.00 |
| Composite-Seen · LoadDishwasher | 1 | 5 | 20% | 542.68 | 898,868.40 | 2,864.80 |
| Composite-Unseen · RecycleBottlesByType | 0 | 5 | 0% | 785.30 | 1,926,543.40 | 4,675.80 |
| Composite-Unseen · GatherTableware | 0 | 5 | 0% | 404.88 | 695,211.80 | 2,549.80 |
| Total | 24 | 50 | 48.00% | 350.02 | 679,429.76 | 2,312.08 |
1.3 RoboTwin
Codex + GPT-5.5 (xhigh) with LingBot-VLA. Ten tasks run in demo_randomized, with five validated scene seeds per task, for 50 episodes.
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| adjust_bottle | 4 | 5 | 80% | 624.46 | 1,158,068.40 | 6,144.20 |
| beat_block_hammer | 4 | 5 | 80% | 891.18 | 3,216,841.00 | 14,468.80 |
| handover_block | 2 | 5 | 40% | 2,369.52 | 7,202,960.80 | 29,643.80 |
| place_empty_cup | 5 | 5 | 100% | 431.60 | 775,905.80 | 4,554.60 |
| place_shoe | 5 | 5 | 100% | 672.50 | 1,501,595.00 | 6,732.80 |
| stack_bowls_three | 3 | 5 | 60% | 1,654.54 | 3,416,140.80 | 12,468.00 |
| blocks_ranking_rgb | 4 | 5 | 80% | 1,581.44 | 4,454,111.00 | 16,199.20 |
| lift_pot | 4 | 5 | 80% | 609.38 | 1,751,153.40 | 6,595.00 |
| dump_bin_bigbin | 3 | 5 | 60% | 1,161.22 | 3,479,640.40 | 13,112.80 |
| open_laptop | 4 | 5 | 80% | 1,693.66 | 3,892,900.20 | 15,953.20 |
| Total | 38 | 50 | 76.00% | 1,168.95 | 3,084,931.68 | 12,587.24 |
2. Installation, Exploration, and Memory Reuse
LIBERO-PRO, RoboCasa365, and RoboTwin were checked through installation, exploration, memory creation, and memory reuse after restarting. Five seeds per environment give 15 reuse episodes. Success rates cover only the reuse stage; exploration runs are recorded separately.
LIBERO-PRO and RoboCasa365 explore with seed 0 and test with seeds 1 through 5. RoboTwin explores in demo_clean and tests in demo_randomized with seeds 100000, 100001, 100003, 100004, and 100005. Runtime and token usage in this table refer to the exploration stage.
| Environment and Task | Workflow | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|---|
| LIBERO-PRO · libero_object_swap · task 0 | Passed | 5 | 5 | 100% | 194.16 | 1,074,213.00 | 3,224.60 |
| RoboCasa365 · OpenDrawer | Passed | 5 | 5 | 100% | 2,293.80 | 1,812,551.00 | 38,699.00 |
| RoboTwin · beat_block_hammer | Passed | 3 | 5 | 60% | 4,267.00 | 5,088,881.00 | 28,691.00 |
RoboDojo completed installation and a single execution check of put_bottles_into_dustbin. This validates installation and the execution workflow; it is not a task-success-rate evaluation.
3. Sampled Paper Reproduction
Using Harness VLA code and Memory, the reproduction tests cover 80 LIBERO-PRO, 50 RoboCasa365, and 50 RoboTwin episodes, totaling 180. Tasks and seed ranges match Section 1. Configurations and results are listed below.
3.1 LIBERO-PRO
Planning model: Claude Opus 4.7 (xhigh). VLA: Pi0.5. Sixteen tasks, 80 episodes. Environment setup and run guide
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| libero_spatial_task · task 0 | 4 | 5 | 80% | 804.10 | 3,189,007.00 | 31,817.00 |
| libero_spatial_task · task 4 | 5 | 5 | 100% | 464.60 | 1,707,359.00 | 16,834.00 |
| libero_spatial_swap · task 0 | 4 | 5 | 80% | 497.50 | 1,877,124.00 | 14,657.00 |
| libero_spatial_swap · task 4 | 4 | 5 | 80% | 534.30 | 1,629,280.00 | 18,174.00 |
| libero_object_task · task 0 | 3 | 5 | 60% | 721.10 | 2,623,604.00 | 18,532.00 |
| libero_object_task · task 4 | 4 | 5 | 80% | 456.40 | 1,706,284.00 | 10,606.00 |
| libero_object_swap · task 0 | 4 | 5 | 80% | 485.40 | 1,565,337.00 | 12,344.00 |
| libero_object_swap · task 4 | 5 | 5 | 100% | 282.40 | 913,677.00 | 10,398.00 |
| libero_goal_task · task 0 | 1 | 5 | 20% | 314.30 | 730,283.00 | 12,251.00 |
| libero_goal_task · task 7 | 5 | 5 | 100% | 119.00 | 362,758.00 | 4,226.00 |
| libero_goal_swap · task 0 | 0 | 5 | 0% | 1,066.90 | 3,140,629.00 | 44,427.00 |
| libero_goal_swap · task 7 | 5 | 5 | 100% | 86.50 | 318,485.00 | 2,915.00 |
| libero_10_task · task 3 | 1 | 5 | 20% | 1,207.30 | 3,659,185.00 | 31,154.00 |
| libero_10_task · task 9 | 0 | 5 | 0% | 841.10 | 1,797,568.00 | 26,700.00 |
| libero_10_swap · task 3 | 0 | 5 | 0% | 904.80 | 2,799,185.00 | 25,717.00 |
| libero_10_swap · task 9 | 0 | 5 | 0% | 376.10 | 896,539.00 | 14,635.00 |
| Total | 45 | 80 | 56.25% | 572.60 | 1,807,269.00 | 18,462.00 |
3.2 RoboCasa365
Planner: Codex + GPT-5.5 (xhigh). VLA: RLDX-1. Seven tasks, 50 episodes. Environment setup and run guide
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| OpenDrawer | 8 | 10 | 80% | 339.40 | 1,360,316.00 | 10,583.00 |
| NavigateKitchen | 10 | 10 | 100% | 288.00 | 1,286,723.00 | 13,702.00 |
| PickPlaceCounterToCabinet | 8 | 10 | 80% | 317.20 | 1,626,763.00 | 10,042.00 |
| PrepareCoffee | 0 | 5 | 0% | 2,435.90 | 9,453,302.00 | 49,694.00 |
| LoadDishwasher | 2 | 5 | 40% | 1,744.40 | 5,114,983.00 | 32,871.00 |
| RecycleBottlesByType | 0 | 5 | 0% | 2,332.60 | 8,435,321.00 | 47,109.00 |
| GatherTableware | 0 | 5 | 0% | 1,760.80 | 7,701,994.00 | 51,364.00 |
| Total | 28 | 50 | 56.00% | 1,016.29 | 3,925,320.40 | 24,969.20 |
3.3 RoboTwin
Planner: Codex + GPT-5.5 (xhigh). VLA: LingBot-VLA. Ten tasks, 50 episodes, in demo_randomized. Environment setup and run guide
| Suite and Task | Succeeded | Episodes | Success Rate | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|---|
| adjust_bottle | 5 | 5 | 100% | 272.40 | 303,276.00 | 1,861.00 |
| beat_block_hammer | 2 | 5 | 40% | 753.10 | 4,075,393.00 | 21,435.00 |
| handover_block | 0 | 5 | 0% | 2,317.00 | 11,461,447.00 | 45,969.00 |
| place_empty_cup | 5 | 5 | 100% | 343.30 | 773,897.00 | 5,453.00 |
| place_shoe | 4 | 5 | 80% | 434.90 | 1,399,450.00 | 8,048.00 |
| stack_bowls_three | 2 | 5 | 40% | 738.50 | 2,084,738.00 | 12,753.00 |
| blocks_ranking_rgb | 3 | 5 | 60% | 1,484.70 | 8,932,639.00 | 37,453.00 |
| lift_pot | 2 | 5 | 40% | 690.40 | 3,016,443.00 | 14,398.00 |
| dump_bin_bigbin | 3 | 5 | 60% | 651.90 | 2,620,781.00 | 11,887.00 |
| open_laptop | 5 | 5 | 100% | 349.70 | 845,135.00 | 5,620.00 |
| Total | 31 | 50 | 62.00% | 803.60 | 3,551,320.00 | 16,488.00 |
Code uses the reproduce/libero, reproduce/robocasa, and reproduce/robotwin branches. Memory uses the Hugging Face dataset’s reproduce/memory branch.
4. Real-World Demo Validation
Dual Franka and YAM completed three tasks: tableware sorting, pouring, and drink-label recognition and sorting. The tableware-sorting task uses one bowl and two plates. Runtime and token usage below refer to a single successful execution of each task.
| Platform and Task | Planner Configuration | Result | Time (s) | Input Tokens | Output Tokens |
|---|---|---|---|---|---|
| Dual Franka · demo01: Tableware Sorting | Codex + GPT-6 Astra (xhigh) | Completed | 600.00 | 2,996,944 | 7,503 |
| Dual Franka · demo03: From Placement to Pouring | GPT-6 Astra (low) | Completed | 1,984.00 | 14,163,904 | 24,159 |
| YAM · demo02: Drink Label Recognition and Sorting | GPT-6 Sol (medium) | Completed | 1,886.85 | 12,026,547 | 38,909 |
See the Leaderboard for full benchmark results and configurations.
What's Changed
- Rename physical agent codename to rpent and structural adjustments by @jx-qiu in #1
- docs: fix documentation issues by @wilburx813 in #2
- docs: add Sphinx and Read the Docs site by @zhexuanxu in #5
- chore: remove libero 10 recipes and memory by @Elessar123 in #9
- update readme by @liushb9 in #11
- doc: add init doc structure by @QuanluZhang in #13
- update readme by @liushb9 in #15
- fix: return libero images as file paths by @wilburx813 in #10
- refactor: refactor dir and clean codes by @qurakchin in #14
- add region-mode cavity-center query to back_project by @Sonorous281 in #6
- doc: update doc by @QuanluZhang in #16
- Restore LIBERO-10 recipes and integrate memory by @Elessar123 in #12
- fix: correct Claude turn accounting by @wilburx813 in #20
- feat: improve install and get started by @andylin-hao in #22
- feat: support different cerebrums in dashboard mode by @wilburx813 in #4
- chore: remove install shell and patch by @andylin-hao in #27
- refactor: rename cerebrum to planner by @Sonorous281 in #28
- Add seed-0 audit JSONs to assist Task-Specific Memory by @Elessar123 in #23
- fix: update spatial swap task 4 recipe by @wilburx813 in #24
- docs: add bilingual Harness VLA publication pages by @huanming99 in #19
- fix: add api image reader by @wilburx813 in #26
- docs: organize content related to robocasa by @Sonorous281 in #32
- raise claude agent sdk stream buffer by @Sonorous281 in #7
- refactor: split cli/main.py and add http/daemon primitives by @qurakchin in #18
- update readme, add link and fix table by @liushb9 in #36
- Fix: correct broken HuggingFace checkpoint URL in README and documentation by @catr1xLiu in #39
- fix: Update renamed LIBERO package name and python version bound by @catr1xLiu in #38
- feat: move all deps to pypi packages by @andylin-hao in #40
- Implement interactive CLI with prompt-toolkit by @jx-qiu in #3
- feat: text-only mode to text models by @andylin-hao in #43
- refactor fetching resources from huggingface by @Sonorous281 in #47
- delete document part and update readme by @liushb9 in #46
- docs: standardize LIBERO-PRO capitalization by @huanming99 in #48
- refactor(envs): move env-specific CLI + runtime init behind EnvSpec by @qurakchin in #41
- Clean up libero prompt organization by @wilburx813 in #35
- feat: integrate SAM3 segmentation into RPent by @wilburx813 in #30
- fix: make parent-watch opt-in to prevent false shutdown on standalone launch by @qurakchin in #55
- docs: improve LIBERO usage and development guides by @wilburx813 in #50
- docs: refactor overview and development pages by @Sonorous281 in #59
- fix: fail fast when a spawned server crashes before ready by @qurakchin in #53
- refactor: consolidate RPC envelope, error handling, and dispatch into shared layer by @qurakchin in #63
- fix: record Claude thinking blocks by @wilburx813 in #58
- fix: export segment calls in libero recipes by @wilburx813 in #54
- fix: separate termination and truncation by @wilburx813 in #61
- fix: align EGL device with --cuda-device to prevent env_server crash on multi-GPU by @qurakchin in #62
- refactor(libero): remove unused primitives and centralize artifact paths by @wilburx813 in #60
- feat(dashboard): improve live monitoring and add Claude interaction by @wilburx813 in #65
- docs: fix cross-reference labels, typos, and backtick errors by @qurakchin in #66
- fix(dashboard): support standard LIBERO suites by @wkzCode in #71
- feat(dashboard): add sequential task session control by @wilburx813 in #69
- feat(dashboard): support API/Codex planners and environment specs by @wilburx813 in #72
- fix(api): handle read_image file errors by @wkzCode in #74
- update readme by @liushb9 in #75
- feat(env): manage env state centrally by @jx-qiu in #73
- docs: quickstart checkpoint downloads by @Sonorous281 in #78
- docs: update Dashboard session documentation by @wilburx813 in #70
- refactor: add env / vla base class by @qurakchin in #88
- fix: add Base prefix to the env base classes by @qurakchin in #90
- RoboCasa365 runtime refactor and install docs by @qurakchin in #81
- docs: add LIBERO reproduction results by @wilburx813 in #91
- Create LICENSE by @QuanluZhang in #95
- feat(robotwin): add robotwin infra integration by @Sonorous281 in #84
- feat(libero): add exploration and layered memory evaluation by @Elessar123 in #79
- fix(libero): pin mujoco to 3.3.0 by @andylin-hao in #100
- refactor: remove codex proxy, add namespace tool support back by @qurakchin in #101
- refactor: rename envs to robots by @qurakchin in #94
- fix(robotwin): dashboard optimization by @Sonorous281 in #103
- refactor(runtime): unify environment runtime initialization by @wilburx813 in #92
- feat(planner): add configurable reasoning effort by @Elessar123 in #102
- refactor: refactor rpc / components path by @qurakchin in #109
- refactor: make robot resources repo id configurable by @Sonorous281 in #106
- refactor: migrate LIBERO RPC adapters to shared base classes by @wilburx813 in #99
- [doc] update readme by @QuanluZhang in #107
- fix: pass reasoning effort only to Codex turns by @wilburx813 in #112
- ci: establish pre-commit baseline by @wilburx813 in #113
- fix(robotwin): initialize native evaluator state after reset by @huanming99 in #104
- fix(robotwin): work around place_fan reward construction bug by @huanming99 in #117
- fix(robotwin): support close-only env teardown by @huanming99 in #119
- Refactor: lift RPC dispatch into RpcFacade base by @qurakchin in #110
- refactor: introduce MemoryManager and reorganize session and prompt by @Sonorous281 in #115
- feat(robotwin): add clean-derived planner guidance by @huanming99 in #105
- fix(robocasa): preserve rendered frame orientation by @qurakchin in #121
- fix(memory): allow root-level markdown leaves in read-only memory by @Sonorous281 in #123
- docs: add RoboTwin reproduction guide by @huanming99 in #122
- build(packaging): split extras by environment by @wilburx813 in #114
- Fix: sam3 rpc by @6UOOON9 in #131
- fix: harden tool result and failure handling by @wilburx813 in #118
- feat(robocasa): complete minimal reproducible integration by @ye1457 in #130
- docs: add robot integration guidelines by @Sonorous281 in #134
- feat: support pi05 for different env by @qurakchin in #136
- test: add offline contract suite and CI workflows by @wilburx813 in #120
- fix(robocasa): enforce live task language for VLA execution by @ye1457 in #140
- fix: improve SAM3 and Codex CLI installation by @JamesHuang0319 in #139
- fix(rpc): bypass proxies for localhost HTTP endpoints by @ye1457 in #144
- fix(robotwin): use mirror for pinned runtime by @Sonorous281 in #145
- feat: support session-aware rpc client / server by @qurakchin in #132
- feat(eval): add extensible run-result finalization by @ye1457 in #148
- refactor(memory): unify cross-robot memory by @Sonorous281 in #142
- feat(robocasa): define and document the target50 evaluation suite by @ye1457 in #141
- fix: serialize concurrent MCP tool calls across planner backends by @qurakchin in #150
- docs(robotwin): clarify reproduction branches by @huanming99 in #149
- build(deps): simplify LIBERO dependency declarations by @wilburx813 in #154
- feat(dashboard): add environment-owned specs and primitive controls by @wilburx813 in #108
- feat(franka): franka single-arm and dual-arm support by @jx-qiu in #98
- fix(franka): remove redundant primitive reset call during toolkit init by @jx-qiu in #161
- feat: combine LLM planning with lightweight execution by @Elessar123 in #111
- test(rpc): relocate dispatch tests under unit_tests by @qurakchin in #163
- test(ci): add embodied GPU end-to-end suites by @wilburx813 in #147
- refactor: unify RLinf python path resolution and to_numpy_tree by @jx-qiu in #162
- feat(libero): add an opt-in data flywheel by @huanming99 in #138
- build(robotwin): converge runtime dependencies by @Sonorous281 in #169
- fix(ci): repair GPU E2E setup and checks by @wilburx813 in #167
- feat(planner): add shared LLM backend connectivity check by @JamesHuang0319 in #152
- fix(robotwin): use requested seed for language prewalk initialization by @huanming99 in #177
- test(robocasa): replace full task-seed matrix with representative smoke checks by @ye1457 in #159
- ci(unittest): add aggregate gate job for a single required check by @qurakchin in #183
- Add dual-Franka real-world deployment, perception, and operator workflows by @NewYork424 in #176
- feat(planner): configure context limits for Codex providers by @huanming99 in #160
- refactor: refactor some components(sam3 / pi05 vla / molmo) and refactor tests of components by @qurakchin in #180
- build(deps): allow uv prereleases for transitive explicit pins by @Sonorous281 in #186
- docs: reorganize tutorials by @huanming99 in #185
- feat(dual-franka): add operator commands and VLA diagnostics by @Code-Hit-ai in #182
- fix(robotwin): bind evaluation language for reproducible results by @huanming99 in #179
- fix(robocasa): pin VLA support resources and document installation checks by @ye1457 in #155
- feat(libero): support all LIBERO-PRO tasks in task cards by @Elessar123 in #171
- docs(benchmarks): add interactive leaderboard and complete paper result tables by @ye1457 in #165
- refactor: prepare runtime naming and independent dashboard cleanup by @wilburx813 in #172
- feat: add rpent shared development skills by @Sonorous281 in #192
- Revert "docs(benchmarks): add interactive leaderboard and complete paper result tables" by @qurakchin in #193
- feat(flash): introduce evaluation-only Flash Mode with selected memory by @Elessar123 in #195
- docs(benchmarks): focus leaderboard on aggregate results by @ye1457 in #194
- docs: add RoboDojo and YAM placeholders to feature matrices by @huanming99 in #197
- ci: remove duplicate documentation build by @wilburx813 in #189
- docs: unify Leaderboard naming and page URL by @ye1457 in #200
- docs(libero): add parallel evaluation guide (zh/en) by @chen00double00guang-cmyk in #191
- docs(leaderboard): add Qwen scores and cost summaries by @ye1457 in #205
- docs: preserve legacy Leaderboard links and fragments by @ye1457 in #204
- fix(session): preserve control-flow exceptions by @silence-breaker in #208
- docs(leaderboard): add GPT-6 motor-only Long Task result by @ye1457 in #207
- refactor(franka): load hand-eye calibration from easy_handeye YAMLs by @jx-qiu in #168
- refactor: port RPent franka code to upstream RLinf by @jx-qiu in #188
- docs(usage): add real-world demo gallery by @Sonorous281 in #211
- docs(usage): split real-world demos by robot by @Sonorous281 in #212
- feat(explore): add minimal RoboCasa and RoboTwin support by @ZhuWeiwen315 in #164
- fix(memory): handle malformed YAML frontmatter by @skyswordx in #209
- docs(readme): add Cosmos Policy to the WAM feature matrix by @Gaojr-2024 in #217
- feat(memory): standardize task layers and enable RoboCasa global memory by @ye1457 in #190
- feat(memory): select versioned LIBERO corpora by model by @ye1457 in #202
- docs(readme): align zh-CN Franka support with English by @mintonight in #213
- docs(leaderboard): synchronize scores and evaluation notes by @ye1457 in #215
- feat(planner): adopt native Pydantic AI runtime by @wilburx813 in #196
- docs(zh): join CJK line wraps in rst_source by @mintonight in #214
- fix(libero): inspect initial state before reading memory by @Sonorous281 in #223
- docs(site): reorganize bilingual docs and unify page styles by @Sonorous281 in #222
- fix(codex): enforce turn budgets for reasoning and tool-only responses by @ye1457 in #201
- refactor(robots): port robots to latest upstream RLinf APIs by @jx-qiu in #216
- feat(yam): integrate yam with RPent by @scilwb in #184
- feat(calibration): add fixed-camera Franka hand-eye calibration by @Code-Hit-ai in #224
- feat(robodojo): add robot backend with eval-fair Flash replay by @littleZ05 in #96
- feat(franka): support Flash replay for single and dual arms by @Code-Hit-ai in #225
- fix(tests): stream offline planner responses by @Travor278 in #229
- refactor(tools): derive native tools from functions and bound methods by @wilburx813 in #198
- feat(robots): add opt-in direct action primitive by @Elessar123 in #203
- docs: clarify module ownership and compatibility guidelines by @Sonorous281 in #247
- chore(release): sync v0.1 with main by @Sonorous281 in #248
- feat(franka): add isolated agent fallback for Flash grounding （pr226） by @Code-Hit-ai in #241
- feat(tools): support parallel readonly calls and coordinated cancellation by @wilburx813 in #199
- refactor(robocasa): consolidate the current Target50 protocol by @ye1457 in #236
- fix(planner): allow reasoning tokens in connectivity probes by @Hiforever-h in #268
- fix(tests): provide global memory for RoboCasa E2E by @wilburx813 in #235
- fix(dual-franka): align RLinf v0.4 deployment and control diagnostics (v0.1) by @NewYork424 in #262
- feat(calibration): preview and exclude collected samples (v0.1) by @NewYork424 in #264
- fix(libero): reject invalid task reset selections by @Travor278 in #232
- feat(dual-franka): support operator input in Dashboard (v0.1) by @NewYork424 in #263
- fix(session): specify H.264 encoder for MP4 frame export by @huanming99 in #277
- feat(memory): select HF sources and replay pinned records(v0.1) by @Travor278 in #230
- fix(libero): hold LIBERO end-effector position during wrist and pitch rotation (v0.1) by @mvanhorn in #240
- fix(libero): honor yaw targets when position converges (v0.1) by @Travor278 in #231
- test(robocasa): move simulator smoke checks into the E2E suite (v0.1) by @ye1457 in #269
- fix(franka): constrain PyArrow for NumPy 1 compatibility (v0.1) by @NewYork424 in #278
- refactor(leaderboard): centralize data and generate documentation assets(v0.1) by @wilburx813 in #245
- docs(robocasa): use the centralized leaderboard (v0.1) by @ye1457 in #283
- fix(planner): forward --reasoning-effort to OpenAI-compatible chat models by @akushonkamen in #281
- feat(memory): select RoboCasa and RoboTwin HF corpora (v0.1) by @Elessar123 in #251
- docs(leaderboard): disclose evaluation dates and VLA training scope (v0.1) by @ye1457 in #261
- fix(robocasa): clarify exploration memory access by @Hiforever-h in #280
- build(robotwin): update runtime dependencies(v0.1) by @Sonorous281 in #287
- docs(install): simplify PyTorch and Molmo setup (v0.1) by @ye1457 in #282
- fix(planner): hand off exploration after truncated reasoning responses(v0.1) by @akushonkamen in #279
- fix(robodojo): constrain PyArrow and streamline setup guide(v0.1) by @wilburx813 in #293
- feat(dual-franka): unify operator reset and manual runtime(v0.1) by @NewYork424 in #284
- fix(molmo): parse image IDs in Molmo2 point coordinates(v0.1) by @Yang-star-code in #275
- docs(dual-franka): streamline deployment and calibration (v0.1) by @NewYork424 in #285
- docs(robocasa): correct Astra reasoning effort to high (v0.1) by @ye1457 in #301
- refactor(yam): port followers to RLinf robot interface (v0.1) by @andylin-hao in #274
- docs: add RPent v0.1 release notes by @Sonorous281 in #300
- chore(release): set package version to 0.1.0 by @Sonorous281 in #308
New Contributors
- @jx-qiu made their first contribution in #1
- @wilburx813 made their first contribution in #2
- @zhexuanxu made their first contribution in #5
- @Elessar123 made their first contribution in #9
- @liushb9 made their first contribution in #11
- @QuanluZhang made their first contribution in #13
- @qurakchin made their first contribution in #14
- @Sonorous281 made their first contribution in #6
- @andylin-hao made their first contribution in #22
- @catr1xLiu made their first contribution in #39
- @wkzCode made their first contribution in #71
- @6UOOON9 made their first contribution in #131
- @JamesHuang0319 made their first contribution in #139
- @Code-Hit-ai made their first contribution in #182
- @chen00double00guang-cmyk made their first contribution in #191
- @silence-breaker made their first contribution in #208
- @ZhuWeiwen315 made their first contribution in #164
- @skyswordx made their first contribution in #209
- @Gaojr-2024 made their first contribution in #217
- @mintonight made their first contribution in #213
- @scilwb made their first contribution in #184
- @littleZ05 made their first contribution in #96
- @Travor278 made their first contribution in #229
- @Hiforever-h made their first contribution in #268
- @mvanhorn made their first contribution in #240
- @akushonkamen made their first contribution in #281
- @Yang-star-code made their first contribution in #275
Additional thanks to @zoeyuchao, @Brunch-Life, and @ucfyao.
Full Changelog: https://github.com/RLinf/RPent/commits/v0.1
