# Alloy Robotics

> Alloy Robotics is a robotics data platform for commercialised robotics teams. Automates mission analysis, enables semantic search across MCAP and ROS2 data, and tracks scenarios over time. Turn weeks of manual analysis into under 1 hour.

## Platform

- [Alloy Platform](https://usealloy.ai): Robotics data platform for commercialised robotics teams
- [Mission Summaries](https://usealloy.ai): Automated mission reports with maps, time-series graphs, and key events in under 1 hour instead of days
- [Scenario Tracking](https://usealloy.ai): Save failure scenarios, auto-detect regressions, track issues over time across software builds

## Core Capabilities

**Automated Mission Summaries**: Generate comprehensive mission reports with maps, graphs, key events, and before/after comparisons. Customer-facing ready reports showing robot performance and impact.

**Scenario Tracking & Observability**: Save specific failure scenarios like GPS errors after sharp maneuvers or emergency braking in rain. Automatically detect when scenarios reappear in new builds and track frequency to identify regressions.

**Semantic Search**: Natural language search across thousands of robot missions without manual labeling. Query using contextual language even when data is unlabeled.

## Target Users

- Validation teams at commercialised robotics companies (autonomous vehicles, drones, underwater robots, maritime systems, defense robotics)
- Robotics researchers conducting experiments and analyzing sensor data
- ML engineers curating training datasets from production robot data
- Product teams reporting on robot performance to customers

## Use Cases

- **Validation Automation**: Analyze position errors across 50 drone flights to validate IMU dead reckoning performance
- **Regression Detection**: Track whether emergency braking scenarios increase or decrease after each software release
- **ML Dataset Curation**: Curate edge case training data like "traffic in rain" without manually reviewing thousands of hours
- **Fleet Scaling**: Scale from 10 robots to 100 robots without 10x-ing headcount

## Data Formats

- MCAP files (robotics industry standard)
- ROS2 bag files
- Multi-modal sensor data: GPS, IMU, cameras, LiDAR, system logs with synchronized timestamps
- No manual labeling required before analysis

## Key Differentiators

- **Time savings**: Week-long analysis workflows reduced to under 1 hour through automation
- **Scale**: Handles terabytes per day, processes 70+ missions daily
- **No labeling required**: Embeddings enable immediate querying of unlabeled data
- **Multi-modal understanding**: Simultaneously searches across images, time-series, and logs
- **Customer-facing reporting**: Automated reports ready to share with customers

## Positioning

Alloy sits above debug/replay tools like Foxglove and Rerun in the workflow. Use Foxglove to visualize individual recordings in detail. Use Alloy to search across thousands of recordings, generate automated reports, and track scenarios over time. Complementary tools used in sequence - Alloy handles curation and analysis, Foxglove handles detailed inspection.

## Open Source Tools

- [tabular2mcap](https://github.com/alloyrobotics/tabular2mcap): Convert CSV, Excel, and Parquet files to MCAP format using pandas. Python tool for robotics teams with tabular data.

## Typical Customer Profile

Companies scaling robot deployments across multiple locations, frustrated by manual effort required to extract trends and insights from large data volumes, currently spending thousands per month on data analysis labor.

## Real-World Applications

- Underwater hull cleaning robots processing hundreds of terabytes
- Autonomous defense systems validation
- IMU/INS hardware manufacturers testing validation units
- Maritime surveying and detection robots on regular missions
- Telesurgery robot validation and performance analysis

## Technical Details

- SOC2 compliant with single-tenant architecture
- Self-onboarding available for company email domains
- Integrates with standard robotics tooling and workflows
- Upload MCAP files from ros2 bag recordings, receive automated reports and searchable mission library

## Contact

- Email: joe@usealloy.ai
- Website: https://usealloy.ai
- GitHub: https://github.com/alloyrobotics