Smart dog function experience

Important Prerequisites

1. The robot dog needs to be connected to a WiFi network. You can refer to the tutorial [1. Basic operations of Raspberry Pi system] --> [6. Connect to WiFi] for detailed operation steps.
2. The robot dog must be configured with the API_KEY related to the large model. You can go to [7. AI large model interaction] --> [1. Prerequisites for using large models] for detailed steps.
3. This feature is an advanced version of embodied intelligence, and requires registration of all platforms mentioned in the configuration used in the large model premise.

Quick Experience

After completing the important prerequisite operation steps, you can quickly get started with the embodied intelligence gameplay. The following are the specific operation steps and example gameplay

1. First, enter the Smart Dog in the sample program, press the button in the lower right corner of the screen to enter this case, and you need to wait about 20 seconds because this case loads a lot of things.

2. The following interface appears, which can be awakened by "lulu"

3. When you hear a "DING" sound, it means it has been awakened and you can carry out complex voice intercom.
4. The following are some specific examples. For example: Example 1: Help me pick up the red ball and put it on the left, then do a dance. Example 2: Follow the blue line and walk forward. Clear any obstacles along the way. Note: For line patrol cases, you need to press the upper right corner of the screen to return to voice communication, otherwise it will continue to patrol. Example 3: Move forward for 2 seconds, show the robot arm, and then perform face detection. Note: For face detection cases, you need to press the upper right corner of the screen to return to voice communication, otherwise it will continue to detect faces. Example 4: Do push-ups, then track the object next to the red ball. Example 5: Help me kick the green ball off the table.

In addition to the simple basic movements of moving forward, backward, left, and right, some basic action groups currently built into this case are shown in the figure below:

The semantics of the command can be modified according to the examples, or combined with the existing action groups to create more command semantics. Notes

• Once you enter the face detection, skeleton detection, license plate detection, line patrol, and object tracking cases, you need to press the button in the upper right corner of the screen to exit the current mode. For example: after face detection, skeleton recognition is performed and then a dance is performed. If you want to enter skeleton recognition after entering face detection, you first press the button in the upper right corner of the screen to exit face detection and then enter skeleton recognition. The same is true if you want to perform a dance.

5. After voice recognition, the dog will then perform corresponding actions based on the semantics of what was said previously.
6. After executing the command, you can repeat steps 2-6 to continue the experience. Press the button in the lower left corner of the dog head screen to exit this example.
7. The flowchart of this function is roughly as shown below: