# microduck_rl | Pollen Robotics | Knowledge Base | Instagit

RL training environments for Microduck (mjlab)

GitHub Stars: 1.1k

Repository: https://github.com/pollen-robotics/microduck_rl

---

## Articles

### [Purpose of the `_servo_joint_ids()` Helper in Microduck RL: How It Isolates Active Joints in Isaac Gym Simulations](/pollen-robotics/microduck_rl/purpose-servo-joint-ids-helper-microduck-rl)

Discover how Microduck RL's _servo_joint_ids() helper isolates active joints, preventing accidental manipulation of physics elements and ensuring accurate reward functions in Isaac Gym simulations.

- Tags: internals
- Published: 2026-09-08

### [How Backlash Is Modeled in the Microduck Robot for Sim2real Transfer](/pollen-robotics/microduck_rl/how-is-backlash-modeled-microduck-robot-sim2real-transfer)

Learn how Microduck models backlash for sim2real transfer using hinge joints, a custom encoder actuator, and environment wrappers. Improve robot control and performance.

- Tags: how-to-guide
- Published: 2026-09-08

### [How Microduck RL Handles NaN Values to Prevent Training Crashes](/pollen-robotics/microduck_rl/how-are-nan-values-handled-prevent-training-crashes-microduck-rl)

Learn how Microduck RL prevents training crashes from NaN values using four mechanisms: reward sanitisation, advantage sanitisation, observation NaN policies, and nan-state termination.

- Tags: how-to-guide
- Published: 2026-09-08

### [Microduck RL MJCF Robot Models: Complete Guide to All 8 XML Configurations](/pollen-robotics/microduck_rl/what-are-robot-mjcf-models-used-microduck-rl)

Explore the 8 Microduck RL MJCF robot models for walking, ground-contact, roller-skate, and backlash variants. Access all configurations in this comprehensive guide.

- Tags: getting-started
- Published: 2026-09-08

### [How to Publish Microduck RL Policies to HuggingFace Hub](/pollen-robotics/microduck_rl/how-to-publish-microduck-rl-policies-huggingface-hub)

Easily publish Microduck RL policies to HuggingFace Hub. Export ONNX, generate manifests, and upload with a single command using uv run publish.

- Tags: how-to-guide
- Published: 2026-09-08

### [How to Export Microduck RL Checkpoints: A Complete Guide to ONNX Conversion](/pollen-robotics/microduck_rl/recommended-way-export-microduck-rl-checkpoints)

Export Microduck RL checkpoints easily with the official CLI script. Convert trained models to ONNX format, embedding the observation normalizer for seamless deployment. Learn how now.

- Tags: how-to-guide
- Published: 2026-09-08

### [How the Observation Normalizer Is Baked Into the ONNX Export for Microduck RL](/pollen-robotics/microduck_rl/how-is-observation-normalizer-baked-onnx-export-microduck-rl)

Learn how Microduck RL bakes the observation normalizer into ONNX export using the EmpiricalNormalization layer for seamless model integration and improved performance.

- Tags: internals
- Published: 2026-09-08

### [Microduck RL Domain Randomization Strategy: How Non‑Accumulating Perturbations Enable Sim‑to‑Real Transfer](/pollen-robotics/microduck_rl/what-is-domain-randomization-strategy-microduck-rl)

Discover the Microduck RL domain randomization strategy. Learn how non-accumulating perturbations ensure effective sim-to-real transfer for robotics.

- Tags: deep-dive
- Published: 2026-09-08

### [How to Train a Microduck RL Policy with Backlash Simulation: Complete Guide](/pollen-robotics/microduck_rl/how-to-train-microduck-rl-policy-backlash-simulation)

Learn to train a Microduck RL policy with backlash simulation. Follow our guide to select tasks, run training, and export your policy for hardware deployment. Get started today!

- Tags: how-to-guide
- Published: 2026-09-08

### [Episodic Trick Tasks in Microduck RL: Forward Roll, Ground Pick, and Ball Kick Explained](/pollen-robotics/microduck_rl/what-are-episodic-trick-tasks-microduck-rl)

Explore Microduck RL's episodic trick tasks like Forward Roll, Ground Pick, and Ball Kick. Learn about these fixed-duration maneuvers for robot learning.

- Tags: tutorial
- Published: 2026-09-08

### [Microduck RL Locomotion Tasks: Complete Task Catalog for Bipedal Robot Training](/pollen-robotics/microduck_rl/what-are-locomotion-tasks-available-microduck-rl)

Explore over 18 Microduck RL locomotion tasks for bipedal robot training, including velocity control, recovery, manipulation, and acrobatics. Train on flat or rough terrain.

- Tags: deep-dive
- Published: 2026-09-08

### [How Backlash Is Simulated in Microduck RL Environments: A Technical Deep-Dive](/pollen-robotics/microduck_rl/how-is-backlash-simulated-microduck-rl-environments)

Learn how backlash is simulated in Microduck RL environments. Discover the three-layer system for modeling mechanical play, actuator behavior, and task configurations for accurate reinforcement learning.

- Tags: deep-dive
- Published: 2026-09-08

### [Core MDP Functions in Microduck RL: Architecture and Key Files Explained](/pollen-robotics/microduck_rl/key-files-mdp-functions-microduck-rl)

Explore the core MDP functions in Microduck RL, including reward terms, state-reset utilities, and joint-index helpers, all located in src/mjlab_microduck/tasks/mdp.py for efficient robot learning.

- Tags: architecture
- Published: 2026-09-08

### [How Unactuated Joints Are Handled in Microduck Robot Models](/pollen-robotics/microduck_rl/how-are-unactuated-joints-handled-microduck-robot-model)

Learn how unactuated joints are managed in the Microduck robot model. The reinforcement learning pipeline focuses on 14 servo joints by excluding passive joints from actuation, observation, and reward.

- Tags: internals
- Published: 2026-09-08

### [What Are the 14 Joints Controlled by the Microduck Robot?](/pollen-robotics/microduck_rl/what-are-the-14-joints-controlled-microduck-robot)

Discover the 14 joints controlled by the Microduck robot, including leg, neck, and head degrees of freedom, all managed by reinforcement learning in the pollen-robotics/microduck_rl repository.

- Tags: getting-started
- Published: 2026-09-08

### [How Runtime Policy Hot‑Swapping Is Enabled in Microduck RL](/pollen-robotics/microduck_rl/how-is-runtime-policy-hot-swapping-enabled-microduck-rl)

Discover how Microduck RL enables runtime policy hot-swapping using an immutable observation contract and a JSON manifest for automated robotctl commands. Learn more.

- Tags: internals
- Published: 2026-09-08

### [Microduck RL Observation Contract: 61-Dimensional Vector Specification](/pollen-robotics/microduck_rl/what-is-observation-contract-microduck-rl-policies)

Understand the Microduck RL observation contract. Discover the standardized 61-dimensional vector for actor and critic networks in the pollen-robotics/microduck_rl repository.

- Tags: api-reference
- Published: 2026-09-08

### [Domain Randomization Techniques Applied to the Microduck Actuator Model](/pollen-robotics/microduck_rl/what-domain-randomization-techniques-applied-actuator-model)

Discover five domain randomization techniques for the Microduck actuator model: PD gain scaling, friction, damping, armature, and mass variations for robust sim-to-real transfer.

- Tags: deep-dive
- Published: 2026-09-08

### [How Microduck RL Simulates Actuator Physics for Dynamixel Servos: Voltage-Controlled BAM Modeling](/pollen-robotics/microduck_rl/how-microduck-rl-simulate-actuator-physics-dynamixel-servos)

Microduck RL accurately simulates Dynamixel servos with voltage-controlled BAM modeling. Achieve high-fidelity sim-to-real transfer by replicating PD control, voltage sag, and friction.

- Tags: deep-dive
- Published: 2026-09-08

### [BAM Actuator Model in Microduck RL: M6 Configuration for XL330 Servos](/pollen-robotics/microduck_rl/what-is-bam-actuator-model-used-microduck-rl)

Explore the BAM actuator model in Microduck RL, featuring M6 configuration for XL330 servos. Experience full-voltage control, friction simulation, and domain randomization.

- Tags: deep-dive
- Published: 2026-09-08

### [How to Deploy Policies on the Real Microduck Robot: A Complete Pipeline Guide](/pollen-robotics/microduck_rl/how-are-policies-deployed-real-microduck-robot)

Deploy policies to the real Microduck robot using a three-stage pipeline. Export to ONNX, publish manifests to Hugging Face Hub, and execute commands on hardware. Learn the complete pipeline guide.

- Tags: how-to-guide
- Published: 2026-09-08

### [How to Export Microduck RL Policies to ONNX](/pollen-robotics/microduck_rl/process-exporting-microduck-rl-policies-onnx)

Export Microduck RL policies to ONNX using the export.py script. This process downloads checkpoints, rebuilds networks with normalization, and outputs a self-contained ONNX file for seamless integration.

- Tags: how-to-guide
- Published: 2026-09-08

### [What Is the Microduck RL Training Repository Based On?](/pollen-robotics/microduck_rl/what-is-microduck-rl-training-repository-based-on)

Explore the Microduck RL training repository. Discover its foundation in MuJoCo Warp, rsl_rl, and a custom BAM actuator model for sim-to-real transfer in legged robotics.

- Tags: getting-started
- Published: 2026-09-08

### [How to Train a Reinforcement Learning Policy for the Microduck Bipedal Robot: A Complete PPO Pipeline](/pollen-robotics/microduck_rl/how-to-train-rl-policy-microduck-bipedal-robot)

Learn how to train a reinforcement learning policy for the Microduck bipedal robot. Explore the complete PPO pipeline for training, exporting, and deploying policies with configurable environments and domain randomization.

- Tags: how-to-guide
- Published: 2026-09-08

### [Where to Find the Hugging Face Jobs Submission Script for Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-hugging-face-jobs-submission-script)

Locate the Hugging Face Jobs submission script for Microduck RL. Find the train_hf.py script and core submission logic in hf_jobs.py within the pollen-robotics/microduck_rl repository.

- Tags: how-to-guide
- Published: 2026-09-02

### [How Microduck RL Achieves Sim‑to‑Real Transfer: A Complete Technical Breakdown](/pollen-robotics/microduck_rl/microduck-rl-achieve-sim2real-transfer)

Discover how Microduck RL achieves reliable sim2real transfer. Learn about physics accurate modelling, domain randomisation, and continuous hardware validation.

- Tags: deep-dive
- Published: 2026-09-02

### [What Is mjlab in the Microduck RL Framework? Core Engine Explained](/pollen-robotics/microduck_rl/role-of-mjlab-microduck-rl-framework)

Discover mjlab, the core engine of Microduck RL. Learn how it powers robot learning with its MuJoCo backend, parallel environments, and PPO implementation for efficient training.

- Tags: deep-dive
- Published: 2026-09-02

### [How to Export Trained Policies From Microduck RL to ONNX: Complete Guide](/pollen-robotics/microduck_rl/export-microduck-rl-policies-onnx)

Easily export trained Microduck RL policies to ONNX using the dedicated export script. Convert RSL-RL policies into portable ONNX models with baked-in normalization and metadata.

- Tags: how-to-guide
- Published: 2026-09-02

### [How to Train Microduck RL Policies Using PPO: A Complete Developer Guide](/pollen-robotics/microduck_rl/how-to-train-microduck-rl-policies-ppo)

Learn to train Microduck RL policies with PPO using this developer guide. Export checkpoints to ONNX for real-robot deployment and accelerate your robotics projects.

- Tags: tutorial
- Published: 2026-09-02

### [How Encoder Biases Are Handled in Microduck RL Domain Randomization](/pollen-robotics/microduck_rl/microduck-rl-encoder-biases-domain-randomization)

Learn how Microduck RL addresses encoder bias in domain randomization. Discover its unique approach to sampling, actor-critic observation, and EMA-based reward for unbiased performance.

- Tags: deep-dive
- Published: 2026-09-02

### [How IMU Orientation Is Randomized for the Microduck Robot: Two-Stage Domain Randomization Explained](/pollen-robotics/microduck_rl/microduck-rl-imu-orientation-randomization)

Discover how Microduck robot IMU orientation is randomized using two-stage domain randomization. Learn about per-environment misalignment and direct MuJoCo randomization for robust simulations.

- Tags: deep-dive
- Published: 2026-09-02

### [What is Armature Randomization in Microduck RL? A Domain Randomization Deep-Dive](/pollen-robotics/microduck_rl/microduck-rl-armature-randomization)

Explore armature randomization in Microduck RL, a domain randomization technique that scales rotor inertia to train robust policies against motor variations.

- Tags: deep-dive
- Published: 2026-09-02

### [How Joint Friction Is Randomized in Microduck RL Training](/pollen-robotics/microduck_rl/microduck-rl-joint-friction-randomization)

Discover how Microduck RL randomizes joint friction during training using a BAM actuator friction-scale hook and curriculum-driven ranges that adapt over time.

- Tags: deep-dive
- Published: 2026-09-02

### [How Mass and Inertia Are Randomized in the Microduck RL Simulation](/pollen-robotics/microduck_rl/microduck-rl-mass-inertia-randomization)

Learn how Microduck RL randomizes robot mass and inertia using pseudo inertia domain randomization. Discover the ±5% scale factor applied at startup for varied simulations.

- Tags: internals
- Published: 2026-09-02

### [How to Configure Head COM Randomization in Microduck RL: A Complete Guide](/pollen-robotics/microduck_rl/microduck-rl-configure-head-com-randomization)

Learn how to configure head COM randomization in Microduck RL. Adjust range and body names for effective episode randomization. Get started with our complete guide.

- Tags: how-to-guide
- Published: 2026-09-02

### [How to Enable and Configure COM Randomization in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-enable-com-randomization)

Learn how to enable and configure COM randomization in Microduck RL. Easily adjust CoM position and body sets with specific ranges for enhanced robot control.

- Tags: how-to-guide
- Published: 2026-09-02

### [Domain Randomization Techniques in Microduck RL: A Complete Technical Guide](/pollen-robotics/microduck_rl/microduck-rl-domain-randomization-techniques)

Explore domain randomization techniques in Microduck RL. Learn how physics, sensor noise, and disturbances enhance sim-to-real transfer for bipedal robots.

- Tags: tutorial
- Published: 2026-09-02

### [How Command Slots Are Handled in the Microduck RL Observation Space](/pollen-robotics/microduck_rl/microduck-rl-command-slots-observation-handling)

Understand how Microduck RL manages command slots within its observation space. Learn about the 13-element command block and stable observation contracts for seamless task execution.

- Tags: how-to-guide
- Published: 2026-09-02

### [MicroDuck 61-Dimensional Observation Space Structure Explained](/pollen-robotics/microduck_rl/microduck-rl-61-dimensional-observation-structure)

Understand the 61-dimensional observation space structure in MicroDuck RL. Learn how it combines angular velocity, gravity, joint states, and actions for policy portability across tasks.

- Tags: deep-dive
- Published: 2026-09-02

### [Microduck RL Observation Contracts: Understanding the 61-Dimensional Policy Interface](/pollen-robotics/microduck_rl/microduck-rl-observation-contracts)

Discover the 61-dimensional observation contract for Microduck RL policies. Learn how proprioceptive sensors and command parameters form the shared actor-critic input.

- Tags: deep-dive
- Published: 2026-09-02

### [How to Define Reward Functions for Microduck RL Tasks](/pollen-robotics/microduck_rl/microduck-rl-define-reward-functions)

Learn how to define Microduck RL reward functions using Python callables and RewardTermCfg objects in task configurations. Master your reinforcement learning tasks.

- Tags: how-to-guide
- Published: 2026-09-02

### [Custom MDP Functions in Microduck RL: A Complete Guide to Robot Learning Rewards and Terminations](/pollen-robotics/microduck_rl/microduck-rl-custom-mdp-functions)

Explore custom MDP functions in Microduck RL. Learn how to define robot learning rewards and terminations using over 60 PyTorch functions in this comprehensive guide.

- Tags: deep-dive
- Published: 2026-09-02

### [How Microduck RL Models Backlash in Its Actuator Architecture: A Complete Technical Guide](/pollen-robotics/microduck_rl/microduck-rl-backlash-actuator-modeling)

Discover how Microduck RL models actuator backlash with passive hinges and routed encoder feedback. A complete technical guide to replicating real hardware behavior.

- Tags: deep-dive
- Published: 2026-09-02

### [How FrictionDRBamActuator Works for Domain Randomization in microduck_rl](/pollen-robotics/microduck_rl/frictiondrbamactuator-domain-randomization)

Learn how FrictionDRBamActuator enhances domain randomization in microduck_rl by injecting friction_scale for non-accumulating friction control. Achieve robust RL training.

- Tags: how-to-guide
- Published: 2026-09-02

### [What Is the BAM Actuator Model in Microduck RL? Understanding the BAM M6 Implementation](/pollen-robotics/microduck_rl/microduck-rl-bam-actuator-model)

Explore the BAM actuator model in Microduck RL. Understand how it emulates Dynamixel servos for robot control and its M6 implementation.

- Tags: deep-dive
- Published: 2026-09-02

### [How Microduck Robot Models Are Defined in MJCF Format](/pollen-robotics/microduck_rl/microduck-robot-models-mjcf-format)

Learn how Microduck robot models are defined in MJCF format using Onshape CAD data and JSON configuration. Generate MJCF XML files for MuJoCo simulation.

- Tags: how-to-guide
- Published: 2026-09-02

### [Core Components of the Microduck RL Architecture: A Complete Technical Guide](/pollen-robotics/microduck_rl/microduck-rl-core-components)

Explore the six core components of the Microduck RL architecture: Environment Configurations, MDP Functions, BAM Actuators, Training CLI, ONNX Export, and Utility Scripts. Master reinforcement learning.

- Tags: deep-dive
- Published: 2026-09-02

### [How to Set Up the Microduck RL Training Framework: Complete Installation and Usage Guide](/pollen-robotics/microduck_rl/how-to-set-up-microduck-rl-training-framework)

Master Microduck RL training with this comprehensive guide. Install and deploy ONNX policies using UV commands and a CUDA GPU. Start your reinforcement learning journey now.

- Tags: how-to-guide
- Published: 2026-09-02

### [What Is the `fallen_too_long` Termination in the VelStand Task?](/pollen-robotics/microduck_rl/microduck-rl-velstand-fallen_too_long-termination-purpose)

Understand the fallen_too_long termination in VelStand tasks. Learn how this backstop prevents zero-cost waiting states during robot recovery training.

- Tags: deep-dive
- Published: 2026-09-01

### [How the fell_over Termination is Handled in the VelStand Task Curriculum](/pollen-robotics/microduck_rl/microduck-rl-velstand-fell_over-termination-curriculum)

Learn how the fell_over termination is managed in the VelStand task curriculum. Discover the strategy for enabling, disabling, and removing this termination condition during training and evaluation.

- Tags: deep-dive
- Published: 2026-09-01

### [VelStand Task in Microduck RL: Combining Walking and Recovery Policies](/pollen-robotics/microduck_rl/microduck-rl-velstand-task-combination)

Discover the VelStand task in Microduck RL. Learn how this unified RL environment combines walking and recovery policies for autonomous locomotion and standing after falls.

- Tags: deep-dive
- Published: 2026-09-01

### [Reward Function for the StandUp Recovery Task in Microduck RL: Architecture and Implementation](/pollen-robotics/microduck_rl/microduck-rl-standup-recovery-reward-function)

Discover the StandUp Recovery reward function in Microduck RL. Learn how Gaussian bonuses and velocity incentives ensure robust standing from a fallen state. Explore the architecture and implementation.

- Tags: architecture
- Published: 2026-09-01

### [Why Turn-in-Place Buckets Are Used in Microduck RL Velocity Walking](/pollen-robotics/microduck_rl/microduck-rl-velocity-walking-turn-in-place-buckets)

Discover why turn-in-place buckets are crucial for Microduck RL's Velocity Walking. Learn how they ensure robust rotation learning by dedicating environments to zero-velocity commands, preventing reliance on rare samples.

- Tags: deep-dive
- Published: 2026-09-01

### [Microduck RL Locomotion Task: Velocity Tracking Environment and Design Decisions](/pollen-robotics/microduck_rl/microduck-rl-main-locomotion-task-design)

Explore the Microduck RL velocity tracking locomotion task. Learn about the bipedal robot's design decisions for following velocity and head-pose commands.

- Tags: deep-dive
- Published: 2026-09-01

### [How Observation and Action Dimensions Change with Backlash Simulation in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-observation-action-dimension-backlash-simulation)

Discover how backlash simulation in Microduck RL preserves 14D observation and action spaces by filtering unactuated DOFs and computing encoder-relative state vectors.

- Tags: deep-dive
- Published: 2026-09-01

### [How to Specify a Backlash Twin Task in Microduck RL Training](/pollen-robotics/microduck_rl/microduck-rl-specify-backlash-twin-task)

Learn how to specify a backlash twin task in Microduck RL training. Microduck RL automatically registers backlash twin tasks during initialization with make backlash variant.

- Tags: how-to-guide
- Published: 2026-09-01

### [How Backlash Is Simulated in the Microduck RL Training Environments](/pollen-robotics/microduck_rl/microduck-rl-backlash-simulation-method)

Learn how backlash is simulated in Microduck RL training environments. Discover techniques for passive hinge joints, backlash-aware observations, and combined joint states for accurate encoder readings.

- Tags: internals
- Published: 2026-09-01

### [How to Perform Joint-Friction Domain Randomization with the BAM Actuator in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-joint-friction-dr-bam-correct-method)

Learn how to perform joint-friction domain randomization with the BAM actuator in Microduck RL. This guide explains effective friction randomization techniques for better simulation results.

- Tags: how-to-guide
- Published: 2026-09-01

### [Why `dof_frictionloss` Is a No‑Op Under the BAM Actuator Model in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-dof_frictionloss-no-op-bam)

Discover why dof_frictionloss is a no-op in Microduck RL's BAM actuator model. Learn how friction is handled internally to prevent double-counting.

- Tags: internals
- Published: 2026-09-01

### [BacklashEncoderBamActuator in Microduck RL: Simulating Real-World Encoder Feedback Through Mechanical Backlash](/pollen-robotics/microduck_rl/microduck-rl-backlashencoderbamactuator-purpose)

Discover how BacklashEncoderBamActuator in Microduck RL simulates real-world encoder feedback. Improve sim-to-real transfer by modeling mechanical backlash for accurate joint angle measurement.

- Tags: how-to-guide
- Published: 2026-09-01

### [How FrictionDRBamActuator Modifies the BAM Actuator Model in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-frictiondrbamactuator-modification)

Learn how FrictionDRBamActuator modifies the BAM actuator model in Microduck RL. Discover non-accumulating domain randomization for sim-to-real transfer.

- Tags: deep-dive
- Published: 2026-09-01

### [BAM M6 Actuator Model for Dynamixel XL330 Servos in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-actuator-model-dynamixel-xl330)

Discover the BAM M6 actuator model powering Dynamixel XL330 servos in Microduck RL. Learn about its voltage-controlled setup and friction characteristics.

- Tags: deep-dive
- Published: 2026-09-01

### [Naming Convention for Unactuated Joints in Microduck RL: The `passive_` Prefix Standard](/pollen-robotics/microduck_rl/microduck-rl-unactuated-joint-naming-convention)

Discover the `passive_` prefix naming convention for unactuated joints in Microduck RL. Learn how it simplifies actuator and observation pipelines for efficient filtering.

- Tags: best-practices
- Published: 2026-09-01

### [Helper Functions for Accessing Servo Joint IDs and Positions in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-servo-joint-id-position-helpers)

Discover Microduck RL helper functions like _servo_joint_ids and _servo_joint_pos to efficiently access servo joint information and isolate actuated degrees of freedom from passive joints.

- Tags: how-to-guide
- Published: 2026-09-01

### [Why Joint Indices Should Not Be Hardcoded in Microduck RL MDP Functions](/pollen-robotics/microduck_rl/microduck-rl-avoid-hardcoding-joint-indices)

Avoid hardcoding joint indices in Microduck RL MDP functions to prevent silent failures and ensure proper action space alignment with your robot's physical configuration.

- Tags: best-practices
- Published: 2026-09-01

### [How the 14 Dynamixel Servos of the Microduck Robot Map to Joint Indices](/pollen-robotics/microduck_rl/microduck-robot-joint-layout-indices)

Discover how the 14 Dynamixel servos in the Microduck robot map to joint indices 0-13. Understand the servo layout for left leg, neck, head, and right leg control.

- Tags: deep-dive
- Published: 2026-09-01

### [Difference Between robot_walk.xml and robot_allcollisions.xml in Microduck RL](/pollen-robotics/microduck_rl/microduck-rl-robot_walk-vs-robot_allcollisions)

Understand the core difference between robot_walk.xml and robot_allcollisions.xml in Microduck RL. Optimize simulations by choosing the right collision geometry for your RL tasks.

- Tags: comparison
- Published: 2026-09-01

### [How to Export Microduck Robot Models from Onshape: A Complete Guide](/pollen-robotics/microduck_rl/microduck-robot-model-export-onshape)

Export Microduck robot models from Onshape using the open-source onshape-to-robot converter to generate MJCF XML and JSON files for MuJoCo.

- Tags: how-to-guide
- Published: 2026-09-01

### [Where Are the MJCF Robot Models Located in Microduck RL? A Complete File Guide](/pollen-robotics/microduck_rl/microduck-rl-mjcf-robot-model-location)

Find MJCF robot models in Microduck RL at src/mjlab_microduck/robot/microduck/. This guide details XML definitions, scene files, JSON configs, and geometry assets for quick access.

- Tags: file-guide
- Published: 2026-09-01

### [Why Microduck RL Zero-Pads Command Slots in Its Observation Space](/pollen-robotics/microduck_rl/microduck-rl-zero-padding-command-slots-purpose)

Microduck RL zero-pads command slots to ensure fixed observation vectors, enabling seamless policy deployment across diverse tasks without recompilation. Learn how.

- Tags: internals
- Published: 2026-09-01

### [How Microduck RL Handles Unused Command Slots in the Observation Space](/pollen-robotics/microduck_rl/microduck-rl-command-slot-handling)

Learn how Microduck RL zero-pads unused command slots in its observation space. Discover how this maintains consistent layouts for policy portability.

- Tags: internals
- Published: 2026-09-01

### [Dimensionality of the Observation Space for the Microduck RL Policy: The 61‑Dimensional Standard](/pollen-robotics/microduck_rl/microduck-rl-observation-space-dimensionality)

Discover the 61-dimensional observation space for the Microduck RL policy. Enable interchangeable ONNX policies across tasks without reshaping tensors.

- Tags: deep-dive
- Published: 2026-09-01