Skip to main content

Multimodal visual understand speech interaction

1. Concept Introduction​

1.1 What is "Visual Understanding"?​

In thelargemodelproject, themultimodal visual understandingfeature enables robots to go beyond simply "seeing" a matrix of pixels and truly "understand" the content, objects, scenes, and relationships within an image. This is like giving robots a pair of thinking eyes.

The core tool for this feature is **seewhat`. When a user issues a command such as "see what's here," the system invokes this tool, triggering a series of background operations that ultimately provide the user with AI-generated analysis of the live image in natural language.

1.2 Brief Implementation Principle​

The basic principle is to feed two different types of information—image (visual information)andtext (linguistic information)—into a powerful multimodal large model (such as LLaVA).

  1. Image Encoding : The model first uses a visual encoder to convert the input image into a computer-interpretable numerical vector. This vector captures image features such as color, shape, and texture.
  2. Text Encoding : Simultaneously, the user's question (e.g., "What's on the table?") is also converted into a text vector.
  3. Cross-modal Fusion : The most critical step is to fuse the image and text vectors in a special "attention layer." Here, the model learns to "focus" on the parts of the image relevant to the question. For example, when asked about the word "table," the model will pay more attention to areas in the image that match the characteristics of a table.
  4. Answer Generation : Finally, a large language model (LLM) generates a descriptive text answer based on this fused information.

Simply put, thisinvolves "highlighting" the corresponding parts of the image with words and then describing the "highlighted" parts with words.

2. Project Architecture​

Key Code​

1. Tool Layer Entry ( largemodel/utils/tools_manager.py )​

Theseewhatfunction in this file defines the tool's execution flow.

# From largemodel/utils/tools_manager.py

class ToolsManager:
# ...

def seewhat(self):
"""
Capture camera frame and analyze environment with AI model.

:return: Dictionary with scene description and image path, or None if failed.
"""
self.node.get_logger().info("Executing seewhat() tool")
image_path = self.capture_frame()
if image_path:
# Use isolated context for image analysis.
analysis_text = self._get_actual_scene_description(image_path)

# Return structured data for the tool chain.
return {
"description": analysis_text,
"image_path": image_path
}
else:
# ... (Error handling)
return None

def _get_actual_scene_description(self, image_path, message_context=None):
"""
Get AI-generated scene description for captured image.

:param image_path: Path to captured image file.
:return: Plain text description of scene.
"""
try:
# ... (Build Prompt)

# Force use of a plain text system prompt with a clean, one-time context.
simple_context = [{
"role": "system",
"content": "You are an image description assistant. ..."
}]

result = self.node.model_client.infer_with_image(image_path, scene_prompt, message=simple_context)
# ... (Processing Result)
return description
except Exception as e:
# ...

2. Model interface layer ( largemodel/utils/large_model_interface.py )​

Theinfer_with_imagefunction in this file is the unified entry point for all image understanding tasks. It is responsible for calling the specific model implementation according to the configuration.

# From largemodel/utils/large_model_interface.py

class model_interface:
# ...
def infer_with_image(self, image_path, text=None, message=None):
"""Unified image inference interface. """
# ... (Prepare message)
try:
# Determine which specific implementation to call based on the value of self.llm_platform
if self.llm_platform == 'ollama':
response_content = self.ollama_infer(self.messages, image_path=image_path)
elif self.llm_platform == 'tongyi':
# ... Logic for calling the Tongyi model
pass
# ... (Logic for other platforms)
# ...
return {'response': response_content, 'messages': self.messages.copy()}

Code Analysis​

This feature's implementation involves two main layers: the tool layer defines the business logic, and the model interface layer is responsible for communicating with the large language model. This layered design is key to achieving platform versatility.

  1. Tool Layer (tools_manager.py):
  • The seewhat function is the core business of the visual understanding function. It encapsulates the entire "seeing" action process: first, it calls the capture_frame method to obtain an image, then calls _get_actual_scene_description to prepare a prompt for the model to analyze the image.
  • The most critical step is calling the infer_with_image method of the model interface layer. It does not care about the underlying model; it only passes the two core data elements, "image" and "analysis instructions," to the model interface layer.
  • Finally, it packages the analysis results (plain text descriptions) received from the model interface layer into a structured dictionary and returns them. This allows upper-layer applications to easily use the analysis results.
  1. Model Interface Layer (large_model_interface.py):
  • The infer_with_image function acts as a "dispatching center." Its primary responsibility is to check the current platform configuration ( self.llm_platform ) and, based on the configuration, dispatch tasks to specific handlers (such as ollama_infer or tongyi_infer ).
  • This layer is key to adapting to different AI platforms. All platform-specific operations (such as data encoding and API call formats) are encapsulated within their respective handlers.
  • In this way, the business logic code in tools_manager.py remains unchanged to support a variety of different large-model backend services. It simply interacts with the unified, stable infer_with_image interface.

In summary, theseewhattool's execution flow demonstrates a clear separation of responsibilities:ToolsManagerdefines the "what" (acquiring an image and requesting analysis), whilemodel_interfacedefines the "how" (selecting the appropriate model platform based on the current configuration and interacting with it). This makes the tutorial's analysis universal; the core code logic remains consistent regardless of whether the user is using the model in online or offline mode. 3. Practical Operations

3.1 Configuring Online LLM​

  1. First, obtain an API key from any of the platforms described in the previous tutorials
  2. Next, update the key in the configuration file. Open the model interface configuration file large_model_interface.yaml: xxxxxxxxxxvim~/hemihex_ws/src/largemodel/config/large_model_interface.yaml
  3. Enter your API Key:Find the corresponding section and paste the API Key you just copied. This example uses the Tongyi Qianwen configuration. xxxxxxxxxx# large_model_interface.yaml## Thousand Questions on Tongyiqianwen_api_key:"sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"# Paste your Keyqianwen_model:"qwen-vl-max-latest"# You can choose the model as needed, such as qwen-turbo, qwen-plus
  4. Open the main configuration file HemiHex.yaml: xxxxxxxxxxvim~/hemihex_ws/src/largemodel/config/HemiHex.yaml
  5. Select the online platform you want to use:Change thellm_platformparameter to the platform name you want to use. xxxxxxxxxx# HemiHex.yamlmodel_service:ros__parameters:# ...llm_platform:'tongyi'#Optional platforms: 'ollama', 'tongyi', 'spark', 'qianfan', 'openrouter'

3.2 Launching and Testing the Functionality​

  1. Start the largemodel main program:Open a terminal and run the following command: xxxxxxxxxxros2 launch largemodel largemodel_control.launch.py
  2. Test:
  • Wake up: Say "Hi,HemiHex" into the microphone.
  • Talk: After the speaker responds, you can say, "What do you see?"
  • Observe the log: In the terminal running thelaunchfile, you should see the following:
    1. The ASR node recognizes your question and prints it.
    2. The model_service node receives the text, calls the LLM, and prints the LLM's response.
  • Listen for the answer: After a while, you should hear the answer from the speaker.