The Lost Dog Mission: Why We Build Thinking Robots

#chapter1 #robotics-intro #perception-planning-control #teen-robotics

> 那个十月的夜晚，白桦树的叶子上结了一层薄霜，我攥着Buddy磨破了毛的网球，坐在林子边的岩石上，嗓子喊哑了也没人应。他是金毛和柯基的混血，最爱的游戏就是追松鼠，那天我们在镇子后山的徒步 trail 上，我停下来拍火红的枫叶，蹲下来系鞋带，一抬头就没了他的影子——他追着一只肥松鼠钻进了密不透风的灌木丛，我再也没能找到他。 > 镇上的志愿搜救队带了两只拉布拉多搜救犬来，可前一天晚上的小雨把他的气味冲得干干净净，Buddy怕生，不会回应陌生人的呼唤，搜到凌晨两点，队长拍了拍我的肩膀说“孩子，我们尽力了”。那时候我根本不知道，三个月后我会站在学校的机器人社团里，盯着一个巴掌大的轮式机器人，下定决心要造一个能把他找回来的机器。 > 这就是所有伟大发明的起点：一个你迫切想解决的问题，加上一个看起来傻得要命的念头。

The Night Buddy Vanished

I still remember the exact moment I realized he was gone. The sun had set an hour earlier, and the only light came from the phone I was holding, its screen glowing with the video I’d taken of Buddy ten minutes prior: he was mid-leap, paws off the ground, chasing a squirrel up an oak tree, his corgi legs kicking up fallen leaves. I’d paused to frame the shot of the maple leaves turning red, then knelt to tie the left boot lace that had come undone on the uphill climb. When I stood up, the trail in front of me was empty. “Buddy!” I yelled, my voice cracking. “Come here, boy!” The wind rustled the birch leaves, and for half a second I thought I heard his collar jingle. I ran into the underbrush, pushing aside low-hanging branches, my sweater catching on thorns. I called his name until my throat burned, until the darkness was so thick I could barely see my own hand in front of my face. I tripped over a tree root and scraped my knee, and that’s when I started crying. I was 16 years old, and I’d lost the dog I’d raised since he was 8 weeks old, the one who slept at the foot of my bed and stole my fries when I wasn’t looking. The search team showed up at 7pm, four volunteers with two yellow Labs trained to find missing people and pets. They worked for four hours, spreading out in a grid, calling his name, letting the dogs sniff his favorite tennis ball I’d brought. But the rain the night before had washed away his scent, and the underbrush was too thick for the dogs to move quickly. One of the volunteers told me later that Buddy was probably too scared to come out, even if he heard us. By 2am, they packed up their gear and left. I sat on that rock for another hour, clutching his tennis ball, until my mom came to pick me up, her face streaked with tears too. I didn’t sleep that night. I scrolled through every photo of him on my phone, watched the video of him chasing the squirrel on loop, and kept replaying the moment I looked up and he was gone. What if he was hurt? What if he was cold? What if he couldn’t find his way back? The search team had done everything they could, but they were limited: they could only search so far, they could only stay out so long, and they couldn’t smell or see through the thick brush the way a dog could. I kept thinking: if only there was a robot that could go into the woods, that didn’t get tired, that could smell his scent, that could see in the dark, that could keep searching long after humans had to go home. That thought stuck with me for the next three days. I went to school with puffy eyes, skipped lunch, and didn’t talk to anyone. My best friend Lila finally had enough on Thursday. “Stop moping,” she said, yanking me out of my chair after physics class. “The robotics club is doing a demo in the gym, you need to come see something.”

The Robot That Changed My Mind

The gym was loud, full of kids laughing and crowding around tables covered in circuit boards, 3D-printed parts, and half-built robots. I didn’t want to be there, but Lila dragged me to the back table where a small, boxy robot with four blue wheels was zipping through a miniature maze made of foam walls. It was no bigger than a textbook, with a small screen on top that glowed green, and a little plastic “head” with two black sensors sticking out like eyes. It moved fast, weaving through the maze’s twists and turns, never bumping into the walls. When it hit a dead end, it didn’t panic or crash – it backed up smoothly, turned exactly 90 degrees, and picked a new path. It never went down the same wrong turn twice, and after 45 seconds, it beeped happily and rolled straight out of the maze’s exit. I stared, my mouth hanging open. “How is it doing that?” I asked. Mr. Chen, the robotics club advisor, was standing next to me. He was a young guy, only a few years out of college, with a faded robotics club hoodie and a 3D-printed keychain shaped like a robot hanging from his belt. “That’s Rover Jr.,” he said, grinning. “He’s built with a Raspberry Pi[^1] for a brain, a camera for eyes, and a few simple sensors to feel his way around. He’s not pre-programmed with the maze path, either – he figures it out as he goes.” “What do you mean, he figures it out?” I asked. “Well, all robots work on three basic principles,” he said, holding up three fingers. “We call them the three pillars of robotics: Perception, Planning, and Control. Let me break it down for you.” He pointed to the little camera on Rover Jr.’s front. “First is Perception. That’s the robot’s way of sensing the world around it, same way you use your eyes, ears, and nose to know where you are and what’s around you. Rover Jr.’s camera ‘sees’ the maze walls, his sensors feel when he’s about to bump into something, and he turns that information into data his brain can understand.” He tapped the Raspberry Pi on top of the robot. “Second is Planning. That’s the brain part. Once he knows where the walls are, he figures out the fastest route to the exit, without going down dead ends. If he hits a wall he didn’t see, he updates his plan on the spot, no human needed to fix it.” He pointed to the robot’s wheels. “Third is Control. That’s the part that makes his body do what his brain tells him to. If he needs to turn left, the control system sends exactly the right amount of power to his wheels to turn smoothly, no wobbly swerving. If he starts to drift off course

The cold October air still clung to my jacket an hour later when Lila dragged me out of the gym, my hands stuffed in my pockets, still clutching Buddy’s chewed-up tennis ball I’d brought just in case. The gym smelled like popcorn and sweat, and kids were still crowding around the robotics tables, oohing and aahing at a 3D-printed dragon that flapped its wings and a tiny drone that zipped through a hoop. I didn’t care about any of it, not until Mr. Chen called us over to the back table where Rover Jr. sat, its little green screen still glowing after finishing the maze. “Wait till you see what it can do,” Mr. Chen said, tapping a few keys on his laptop. Rover Jr.’s little plastic head swiveled toward us, its black sensor eyes blinking, and it beeped once, like it was saying hello. I knelt down to get a closer look. Its body was made of laser-cut acrylic, held together with tiny screws, and the four blue wheels had rubber treads that looked like they could grip even wet leaves. The Raspberry Pi[^1] on top was smaller than my phone, with a tiny cooling fan whirring softly. “It’s not pre-programmed with the maze path, remember?” Mr. Chen said, grinning. “Watch this.” He pulled a new maze layout out of his bag, one I’d never seen before, with more twists and dead ends than the first one. He set Rover Jr. at the start, and it took off immediately, weaving through the new maze, backing out of dead ends, turning exactly 90 degrees every time, no hesitation. It finished in 38 seconds, beeped happily, and rolled right up to my shoe, nudging it gently with its front bumper. I laughed before I could stop myself. “It’s so cute,” I said, reaching out to pat its top. The plastic was warm from the Raspberry Pi running. “Cute, but also really useful,” Mr. Chen said. “Remember I told you about the three pillars? Let me show you how they work, using Rover Jr. as an example.” He held up three fingers again, like he was teaching a middle school science class. “First, Perception. That’s the robot’s way of sensing the world, same way you use your eyes, ears, and nose to know where you are and what’s around you. Rover Jr.’s camera ‘se