Want us to integrate Almond Bot into your workflow?

Let our team handle the integration and programming for your specific use case. Contact us to get started.

​
tldr;

We’ve integrated with the best hardware, vision, and compute with the most popular software libraries to make automation a breeze. Access waypoints, AprilTags, object detection, end-to-end AI and more with a single API.

With Almond Bot, you can:

  • Program basic tasks with waypoints
  • Detect & move relative to AprilTags
  • Detect & classify objects
  • Automate complex tasks with AI

​
Bot API

Our API exposes low and high level control of Almond Bot from specifying joint angles and cartesian pose to tele-op to running end-to-end AI models. Take a look at a few examples below to see what is possible with just a few lines of code.

​
Examples

​
Basic Automation

Move to set points.

from almond.client import AlmondBotClient, Pose
bot = AlmondBotClient()
await bot.connect()

await bot.open_tool()
await bot.set_tool_pose(Pose(x=0.1, y=0.2, z=0.3, roll=0, pitch=0, yaw=0))
await bot.close_tool()
await bot.set_tool_pose(Pose(z=10), is_offset=True)

​
AprilTag

Align with tags around your factory, and then perform relative movements.

from almond.client import AlmondBotClient, Pose
bot = AlmondBotClient()
await bot.connect()

await bot.align_with_apriltag(id=0, size=100)
await bot.open_tool()
await bot.set_tool_pose(Pose(z=10), is_offset=True)
await bot.close_tool()
await bot.move_arc(r=500, Pose(x=-100, y=-100), is_offset=True)
success = await bot.verify_scene("door is open")

​
Object Detection

Contact us to build a custom object detection model for your use case.

from almond.client import AlmondBotClient, Pose
bot = AlmondBotClient()
await bot.connect()

poses = await bot.detect_poses("widgets")
for pose in poses:
    await bot.open_tool()
    await bot.set_tool_pose(pose, is_offset=True)
    await bot.close_tool()
    await bot.set_tool_pose(Pose(x=100, y=100, z=50))
    await bot.open_tool()
success = await bot.verify_scene("all widgets in box")

​
AI

Develop AI directly with our LeRobot fork or use our simple APIs to collect data, train, and run models.

​
Collect Data

from almond.client import AlmondBotClient, Mode
bot = AlmondBotClient()
await bot.connect()

# Enable robot teleop
await bot.set_mode(Mode.TELEOPERATION)

await bot.record_episode(
    task_name="pickup_widget",
    duration_seconds=60
)

​
Train Model

from almond.client import AlmondBotClient, AIModel
bot = AlmondBotClient()
await bot.connect()

training = await bot.train_task(
    task_name="pickup_widget",
    training_name="my_pi0_training",
    model=AIModel.PI0
)

​
Run Model

from almond.client import AlmondBotClient
bot = AlmondBotClient()
await bot.connect()

await bot.run_task(
    task_name="pickup_widget",
    training_name="my_pi0_training"
)

success = await bot.verify_scene(
    "the widget is in the container"
)