lab-robot v0.2.0

What’s new

• Add a configurable direct OpenAI Responses API vision provider (VISION_PROVIDER=openai).
• Send camera JPEG frames and questions directly to the default gpt-4o model; preserve the Xiaozhi vision gateway as the default provider.
• Add mocked API tests for image payloads, model configuration, response validation, missing credentials, HTTP failures, and secret redaction.

Docker

docker pull ghcr.io/zbbio/lab-robot:v0.2.0

For GPT-4o, set VISION_PROVIDER=openai, OPENAI_API_KEY, and optionally OPENAI_MODEL in the local .env file, then recreate the Compose service.

The OpenAI provider was validated with mocked API responses; configure a valid OpenAI API key to make a live request. See the README for setup details.

lab-robot v0.1.0

首次公开发布 lab-robot：基于 MCP 的实验室视觉机器人助手。

• 包含 Python/Node.js 摄像头工具、小智桥接和固定版本的 Opentrons MCP。
• 支持 ONVIF/RTSP、截图、云端 VLM 图像问答及可选 PTZ。
• 提供 Docker Compose、配置模板、运行测试和演示视频。
• ESP32 / ESP-IDF 语音终端属于整机架构；本版本不包含其固件或机械设计文件。

镜像：ghcr.io/zbbio/lab-robot:v0.1.0，由版本工作流验证通过后发布。

部署步骤见 README；真实设备、云端认证与实验协议需在目标环境配置和联调。