Loco Pik

World View

Loco Pik is a group of small alien-like companions from a distant planet. They quietly land on Earth and begin to live on human desks. They are not cold machines or traditional functional robots. Instead, they are soft, curious, and emotional little creatures that accompany people in daily life.

Loco Pik does not mainly communicate through human language. It senses the surrounding environment and expresses itself through small body movements, light, sound, and simple emotional reactions. When it sees something interesting, it may become curious. When it reaches the edge of the table, it becomes nervous and moves backward. When there is no interaction for a long time, it may slowly enter a sleepy state.

In this fictional world, Loco Pik does not have a fixed body shell. Its outer shell can change according to different environments, materials, and emotional states. Some shells may look like stone, wood, ceramic, moss, or other natural textures. This setting explains why the robot can have replaceable shells and why AI-generated forms and 3D printing can be used to create different bodies for the same internal system.

Research

The core idea of Loco Pik is language-free emotional companionship. Instead of designing it as a voice assistant that only answers questions, I want it to feel like a small living creature on the desk. Its emotional interaction is built from several physical outputs: leg movement, ambient light, and short sound feedback. These outputs allow the robot to show different states such as happy, curious, nervous, sleepy, and idle.

The robot will use the Seeed Studio XIAO ESP32S3 Sense as the main controller. This board provides Wi-Fi, BLE, an onboard camera, and an onboard digital microphone. The camera can be used for simple visual sensing, such as detecting objects on the desk. The microphone is kept as a sound input for future voice interaction. For the first working prototype, the goal is not to run a complete large language model inside the robot. Instead, the focus is to build a stable physical system that can sense the environment, move safely, and express emotion through simple behaviors.

The outer shell will be fabricated mainly through 3D printing. By combining AI-generated form exploration with 3D modeling and 3D printing, the robot can have a soft, organic, and non-mechanical appearance. The current front design uses two sensing holes: one for the camera and one for the microphone sound input. A bottom LED will be used as an ambient emotional light, so the front face can remain simple and mysterious.

First Sketch

This is my first concept sketch, along with AI-generated conceptual product renderings based on it. These early images helped me explore the personality, body proportion, surface texture, and interaction direction of Loco Pik. At this stage, the design still included many possible technologies, such as AI dialogue, NFC interaction, emotional sound, replaceable shells, and body movement.

What I Keep From the Original Concept

The world view of Loco Pik is still the foundation of the project. Loco Pik is still a group of small alien companions from a distant planet. They are not designed as cold machines, but as warm and curious desktop creatures. The organic body, the replaceable shell, the alien identity, and the language-free emotional interaction are all kept.

The project also keeps the idea of using AI-generated visual exploration and 3D printing to create natural shells. The shell can be inspired by wood, stone, ceramic, moss, or other natural materials. This makes the robot less mechanical and more like a small creature with its own personality.

What I Reduced or Moved to Later Versions

Some early functions were interesting, but they would make the first prototype too complex. At the midterm stage, I decided to reduce them or move them to future versions.

What I Added or Clarified

During the midterm development, I clarified the hardware system and separated the project into smaller functional parts. Instead of treating the robot as one complex system, I divided it into sensing, movement, sound, light, power, and structure. This helped me decide which parts should be kept in the first prototype and which parts should be moved to future versions.

I first made a hand-drawn wiring plan to understand how the XIAO ESP32S3 Sense, PCA9685 servo driver, servos, ToF sensor, camera, speaker amplifier, LED, NFC module, and battery could be connected together. This sketch helped me check the relationship between different modules before making the PCB. In the later simplified version, I removed the TCRT5000 sensor and kept only one ToF distance sensor for obstacle and edge-distance testing.

Connection With Previous Weeks

The current version of Loco Pik is not an isolated final project idea. It is gradually developed from the skills and experiments I completed in previous Fab Academy weeks. Each weekly assignment helped me test one part of the final system, including electronics design, electronics production, output devices, camera-based AI recognition, and digital fabrication.

Through these weekly experiments, I started to understand how the final robot should be divided into smaller technical systems: structure, movement, sensing, output, power, and interaction. The midterm stage helped me connect these separate tests into one clearer final project plan.

System Diagram and Hardware Architecture

This system diagram shows how the main controller, sensing modules, output devices, movement system, and power system are connected in the first prototype of Loco Pik. The updated hardware system is organized around the XIAO ESP32S3 Sense. The camera and microphone are used as the front sensing organs. The PCA9685 servo driver controls four MG90S servos through the I2C bus. The ToF sensor, optional PN532 NFC module, and PCA9685 can share the same I2C bus. The MAX98357A amplifier uses I2S for sound output, while the WS2812B LED provides bottom ambient light.

The power system is separated from the XIAO logic power. The servos and audio amplifier will be powered by an external 5V or 6V UBEC, while the XIAO only sends control signals. All GND lines must be connected together.

First Version of the Final PCB

Based on the electronics design experience from Week 06, I designed the first version of the PCB for the final project. This PCB is not the complete final control system yet, but it is an important step toward integrating the robot's hardware into a cleaner structure.

The first PCB focuses on organizing the basic connections for the XIAO board, servo connectors, LED output, power input, and extra pin headers. It helps reduce messy jumper wires and makes the robot easier to assemble inside the small 3D printed body.

After milling the board, I soldered the pin headers and checked the basic connection layout. This version is mainly used to verify the physical arrangement of the connectors and the feasibility of the single-sided PCB. From this test, I can further adjust the board size, connector direction, and power routing for the next version.

Mechanical Structure

The robot will use a round 3D printed body. The front shell has two holes. One hole is used for the camera, and the other one is used as a microphone sound hole. This creates a simple alien-like face while keeping the sensing functions visible.

The current version has four servo-driven legs. The four MG90S servos will be mounted around the lower body to create simple walking, turning, and emotional movement. The legs are not designed for fast walking, but for small desktop movements and expressive body gestures.

The camera should be placed close to the shell opening. If the lens is too far from the hole, the shell may block the camera view. Therefore, I will either place the XIAO ESP32S3 Sense close to the front shell or test a short 24Pin 0.5mm FFC/FPC camera extension cable. If the camera extension becomes unstable, I will return to the safer structure where the whole XIAO ESP32S3 Sense board is mounted near the front face.

The bottom of the robot will contain the battery, UBEC, speaker, ToF distance sensor, and bottom ambient LED. In this version, I kept only one ToF distance sensor for obstacle and edge-distance testing.

Updated Behavior Plan

The robot will not rely on a large AI model to control every movement. Instead, it will use a behavior-based system. The camera and ToF distance sensor provide input, and the program decides the emotional state and movement response.

Tasks to Be Completed

About AI Dialogue

For the AI dialogue part, I confirmed that I will use xiaozhi-esp32 as the main technical reference for future development. xiaozhi-esp32 is an open-source ESP32-based AI chatbot project built around MCP. It uses ESP32 hardware as a voice interaction terminal and connects the device to large model capabilities through network communication.

According to the project documentation, xiaozhi-esp32 supports Wi-Fi, offline voice wake-up, WebSocket or MQTT + UDP communication, OPUS audio codec, and a streaming ASR + LLM + TTS voice interaction architecture. It also supports ESP32-C3, ESP32-S3, and ESP32-P4 platforms, and can control device-side modules such as speakers, LEDs, servos, and GPIO through MCP.

This is useful for Loco Pik because the robot also needs to connect AI dialogue with physical behaviors. In the first prototype, I will still focus on local sensing, movement, emotional sound, and bottom light feedback. After the basic robot system becomes stable, xiaozhi-esp32 can be used as the next step to add voice conversation, wake-up interaction, and AI-controlled emotional behaviors.

Updated Materials List / Midterm BOM

This is the midterm version of the bill of materials. At this stage, all components are planned to be sourced from Taobao. The prices are estimated midterm values and will be updated in the final documentation after actual purchasing, fabrication, and system integration.