by Rahul Bhagat
Martin walked back to the cruiser. How was he going to find this kid? He sat down inside the vehicle and browsed through folders containing investigation documents. There was nothing, except for that three-dimensional scan of the van. The van was a rental, and they didn’t have any information on the kid because it had been paid for through ZCash, an anonymous currency.
Now what? Martin looked outside and wondered. Then he noticed something attached to Mr. Wong’s house, and he went back there as fast as he could with the cane.
“I’ll need access to your old video footage,” Martin said to Wang and pointed at the surveillance camera.
Within half an hour, he had a picture of the kid with droopy eyes. He went back to the cruiser and activated SSAI, Suspect Search AI.
“SSAI, find me this person,” Martin said and showed the kid’s image to AI’s camera.
Within a minute, SSAI had completed the task. “The suspect is Jacob L. Smith,” it said.
“Can you find out his current whereabouts?”
“I can say with ninety percent confidence that he is at his parents’ place,” SSAI said.
“How did you find out?” Martin asked.
“It was easy,” the SSAI said. “I started running face search through social media databases for young males in and around DC and found several hits. Then I built an activity map of shortlisted individuals—things like their movement and spending pattern; their appearance in surveillance videos at shopping malls and transit stops, and soon I was able to narrow down to our Mr. Smith. His last transaction was a one-way ticket to Edgewater, Maryland, where his parents live.”
TWENTY-EIGHT
MARTIN TOOK A private, more spacious AV for the journey. The trip to Edgewater was going to be long, and his vehicle was infinitely more comfortable than a police cruiser. As usual, the vehicle was traveling in a tunnel. Martin looked out the window and waited for the grey concrete walls to be replaced by wide-open fields and blue skies. He looked forward to warm rays of the sun on his face.
When automobiles became ubiquitous in the early-twentieth century and started replacing horse-drawn carriages, they brought significant change to the roadways. Police were brought in to manage the flow of vehicles and were eventually replaced by traffic lights, and pedestrian crossings were introduced to reduce accidents and fatalities.
Similarly, the latter part of the twenty-first century saw another significant overhaul, driven mainly by widespread adoption of autonomous vehicles. Smooth-flowing traffic would come to a screeching halt if a pedestrian stepped on the street. And God forbid if there was a fatality. There were cases where the cascading effect of an accident caused traffic disruption hundreds of miles away from the site. Slowly but steadily, cities walled off their streets, and soon, most were moved either underground or overhead, at least in the metropolitan areas.
And massive, multi-lane highways that snaked through city cores disappeared. Citizens and planners labeled them an eyesore and demanded that portly highways be trimmed. In the age of autonomous vehicles, people needed only one lane, or perhaps two for peak traffic hours.
Martin’s vehicle emerged from a tunnel and traveled south on the interstate. It was a real treat to look at the scenery flashing by. After about an hour, the vehicle stopped in front of a modest home, an old home, probably built toward the end of the previous century. The house backed onto the South River, which emptied into the Chesapeake Bay.
An old man answered the door.
“Is Jacob home?” Martin asked, guessing the man was Jacob’s father.
“Yes, he is. What’s your business?” he asked with arched eyebrows.
“I’m Detective Stump from the New Town Police Department. I need to speak with him,” Martin said.
The color drained from the father’s face. “Is he in any trouble?” he asked and stepped aside to let him in.
Martin made his way through the narrow passage. The living room was dark; heavy drapes blocked the sunlight. It was crowded with furniture and knickknacks, and the ceiling was really low. It was claustrophobic in there.
While Martin waited, the father disappeared inside. Soon he was back with his wife and son in tow, all looking very worried. Martin recognized Jacob immediately from his signature droopy eyes, which he had inherited from his mother.
“Remember me?” Martin asked.
The boy nodded but said nothing. He sat down on the couch between his parents.
“I have some questions about the van you rented,” Martin asked. “Why did you rent that vehicle?”
“For a job,” Jacob said, his eyes fixed on the floor.
“Job?”
“I needed money to make rent and saw this ad,” he said.
“What ad?” Martin softened his voice.
“They wanted someone with a driver’s license.” Jacob looked up at him.
“And you have one?” Martin said with surprise. “Why on earth would you need a driver’s license? Those are not easy to acquire.”
A person with a driver’s license had the authority to override traffic rules. That person could overtake other vehicles, switch lanes at will, and control speed and movement of an AV, all without traffic sentinels raising an eyebrow.
“And in my father’s days, everyone and their dog had a driver’s license.” Jacob’s dad tried to lighten the mood.
“I used to work for a towing company,” Jacob said. “I didn’t like the job, so I quit, but I still have the license.”
“I see. And what was this ad about?” Martin asked.
“Rent a van and ride it on the highway.” Jacob started sweating profusely.
“Just ride aimlessly on the highway.”
“No, ride in front of that black AV.”
“The one that crashed?”
Jacob nodded. He didn’t meet Martin’s eyes.
“And then what?”
“Tap the button on the controller,” Jacob said.
“What controller?” Martin was confused now. “Rewind, tell me from the beginning. What happened after you responded to that ad?”
Jacob swallowed and started speaking. “I received a response, with instructions to rent a van, and five thousand dollars.”
“Who was this person?” Martin asked.
“I don’t know. We were chatting on TORChat; it’s anonymous. He told me to leave the van in the neighborhood strip mall and send him the van’s access code,” Jacob said and quickly added, “I was really nervous, but the money was so good.”
Martin nodded. “It always is.”
“The next day, he told me to pick up the van from the parking lot and asked me to wait,” Jacob said.
“Wait?” Martin frowned.
“Yes, and there was something mounted on the back of the van.”
Martin remembered the marks. He sat up straight. “Describe it.”
“It was like a large bulging plate, nothing more. And there was a touchscreen controller inside the van.”
“What did you do after that?”
“I waited. He told me to be ready to drive at a moment’s notice. I waited for three days without a word from him. And then, on the fourth day, I received a message in the morning to be ready,” Jacob said. “I waited all day, and around evening, just before nightfall, he asked me to drive and wait near the ramp. The weather was bad, cloudy and raining. Then he messaged me to get on the highway and speed up. The controller became active and displayed an aerial view of the highway.”
“From a drone?” Martin asked.
“Yes,” Jacob said. “He messaged me to catch up to the black AV. I established contact with the traffic sentinel and increased the speed of the van. Once I caught up, he instructed me to insert the van just ahead of the AV and told me to do so before we arrived at the tunnel.”
“Why?”
“I don’t know. He just told me to do it and tap the red button on the controller immediately after entering the tunnel. It probably had some preprogrammed instructions on it.”
“Last instructions were to remove it and put it in the garbage,” Jacob said.
“Oh no.” Martin wasn’t looking forward to rummaging through a recycling plant. “Where did you dispose of it?”
“I threw it in a ditch,” Jacob said.
Martin perked up. “Good boy! It’s probably the only time in my life that I’m approving of littering.”
TWENTY-NINE
IN A DARK lab, Martin and Natalie huddled around a wide table illuminated by hooded lights suspended overhead. The bulging disk Martin had found in the ditch lay open, its innards spilled across the table. There were other pieces, too, including the controller and parts of a plastic mount that was used to attach the disk to the back of the van.
“This is a murder, Martin, no doubt about it,” Natalie said, twirling a pencil in her hand. “I think I know how this crash was engineered.”
Martin breathed heavily and tightened his jaws. “How?”
“You see this?” She pointed at the open disk with her pencil. “This is a lidar-spoofing device.”
“A what? What is this thing again?”
“Lidar is like radar, but instead of radio waves, it uses light, laser, to detect its surroundings. Basically, the device sends out tiny pulses of light and focuses on the spot where the beam was sent, to catch its reflection. The time it takes for the reflection to come back is how lidar calculates the distance of the object hit by the laser. And it does so thousands of times, rapidly, to paint a picture of its surroundings as a cloud of dots of reflected light.”
Martin nodded. He just wanted to know what a lidar was and had completely forgotten Natalie’s propensity to describe everything in detail. “What does the spoofing device do?” he asked.
“It messes up the picture that lidar builds. The device first reads the vehicle’s lidar pulse and synchronizes with it. It then transmits signals that look just like lidar reflections. If it sent out signals of reflection from a wall, the AV would think there was a wall in front of it. Basically, the spoofing device creates fake images.”
Martin arched his eyebrows and pointed at the disk. “This thing created fake images in front of the AV and caused it to jump in the ravine?”
“Yes, but it’s not that simple; there is more to it. AVs are designed to protect against spoofs, and they have a lot of built-in redundancy.”
“I’m not following you.” Martin sighed. A part of his brain was clamoring for the person who had done this, who had taken Julie away from him.
“You see,” Natalie said, “lidar is only one of many eyes that an AV uses to see its surroundings. Besides lidar, there is radar and camera. If one device fails or is compromised, the AV can check its other eyes to confirm what’s real. Theoretically, the AV should have been able to override this spoof.”
“But it didn’t.”
“No, and I think I know why,” Natalie said, twirling the pencil and sounding a little cocky. “We built in redundancy because each device has its own shortcomings. For example, radar—they are great, can see through rain, fog, snow, even around corners, but in confined places like a tunnel, they are useless. They create a lot of noise; signals bouncing off the walls confuse the sensors. In situations like that, AVs ignore images from radar and rely on onboard cameras and lidar. Same story with cameras. At nighttime or in dark places, they become useless, so in that case, the AV relies on radar and lidar.”
“And lidar? Where does that fail?” Martin asked.
“In bad weather,” Natalie said. “In rain, dust, snow, lidar goes crazy. Because light pulse reflects off snowflakes and dust particles, it creates a cloud that makes no sense to the vehicle’s controller.”
Martin nodded and scratched his chin. This was a lot of information to digest.
Natalie continued. “They all have their strengths and weaknesses, but together, they work very well. So, even if someone uses a lidar-spoofing device, an AV should be able to easily compare its images with those from radar or camera and defeat the spoof. It’s not easy.”
“What happened then? Why did it work in this case?” Martin asked with frustration in his voice.
“Whoever did this planned it very carefully,” Natalie said. “That person found a vulnerable spot and exploited it.”
“How?”
Natalie took a moment to gather her thoughts. “Here’s what I think happened. When the AV entered the tunnel, the signal from radar started bouncing off the walls of the tunnel. That created too much noise, so guess what the AV did?”
“It ignored radar signals,” Martin said.
“Exactly,” Natalie said. “Next, camera. You saw how poor the lighting was in the tunnel; all that grime and dust on the casings. The signal from the camera was downgraded too and ignored. Inside the tunnel, the AV was almost exclusively relying on lidar for navigation. Remember, Jacob activated the spoofing device shortly after entering the tunnel.”
Martin nodded. It was starting to make sense.
“Once the spoofing device had calibrated the signal, it mounted an attack. And guess what the black AV did.”
“It braked,” Martin said, chewing his lower lip.
“Right.” Natalie tapped the pencil on the table. “I think the spoofing device sent out a signal that the van had slowed down, but in truth, it had maintained its speed. The AV braked and slowed to avoid a collision, and this created a wider gap between the vehicles. But the AV never filled that gap to minimize air drag.”
“That was our first clue,” Martin said.
“My theory is that the spoofing device sent signals that made the AV believe it was farther back in the tunnel, and the van was at an optimal distance. That’s why it didn’t fill the gap. Since the AV was ignoring conflicting information from the radar and camera, there was no way for it to cross-check. The controller became confused because the prior map stopped matching the signal from the lidar. Therefore, it moved the map back by a second to match the view from the lidar. You see what’s happening here?”
“Not really,” Martin said with a frown on his face.
“The AV was traveling at two hundred miles per hour, so in one second, it covered 0.06 miles—about thirty feet. The spoofing device was projecting images of the tunnel from thirty feet behind, which made the AV move back its prior map by the same distance.”
“I see. So the vehicle was thinking it was thirty feet behind where it actually was.”
“Exactly.”
A light bulb went off in Martin’s head. “Got it,” he said. “When the AV emerged from the tunnel, it was thinking it still had thirty feet of ground to cover before the tunnel ended, and the road took a sharp left turn. That’s why it kept going straight.”
“You got it, Martin,” Natalie said and thumped the table.
“But the vehicle did brake before it hit the guardrail.”
“That’s not hard to explain,” Natalie said. “Once it emerged from the tunnel, the radar and camera signal quality improved immediately, and the controller realized the lidar view didn’t match what its other two eyes were seeing, and worse, the vehicle was about to hit a guardrail. That’s why it braked desperately.”
“Makes sense…”
“And by that time, the van, too, had turned left and was out of sight. So the ghost images from the spoofing device were gone. It was as if the AV suddenly woke up, realized what was happening, and tried desperately to save its occupants.”
THIRTY
MARTIN AND NATALIE emerged from the dark lab into a bright corridor. Sun poured in through large windows, and for a moment, Martin couldn’t see anything, but then his eyes adjusted. Outside, manicured lawn and rows of flower beds looked resplendent in the midday sun.
“This is a nice place,” Martin said.
“I was really lucky to get an office at the university. Most of my colleagues are at the new campus, in that glass tower,” Nat
alie said.
Martin followed Natalie as they navigated crisscrossing corridors on their way to the cafeteria.
“Any ideas on how we find that mystery person? The guy who hired Jacob?” Natalie asked as they entered the cafeteria.
“Maybe that spoofing device?” Martin said. “Can we trace it back to the person who bought it?”
“I don’t know,” Natalie said absentmindedly while she decided between a tuna and an egg sandwich. “That device was assembled; someone bought parts and put it together.”
“Then let’s trace the parts.”
“Fair enough. Let’s do it.”
After picking up their lunch, they returned to the lab. Natalie methodically examined each part of the spoofing device and noted down its serial number. Martin fed the numbers to AI Sleuth, tracing software, which checked where they originated from. All the pieces pointed to an online electronics retailer that sold DIY kits. They called the company and, to their dismay, learned that the retailer had bought the components in bulk. And it didn’t track which piece went into which kit.
Discouraged, they decided to call it a day. It was getting late, anyway, so they went back to the cafeteria for dinner.
“So what do we do now?” Natalie asked from across the cheap pseudo-wooden table. She picked at her food with disinterest.
“We’ll have to find a way. I’m sure it’s lurking somewhere out there,” Martin said and tried to force a smile. He felt depressed himself and didn’t know what else to do. They ate in silence for a while.
“Let’s wrap up and go home,” Natalie said, pushing away from her barely touched plate.
“We’ll start tomorrow with a fresh mind.” Martin tried to sound cheerful.
Back in the lab, they started reassembling the spoofing device for storage.
Martin picked up a small piece of broken plastic. “Where did this come from?” he asked.
“It’s part of the clamp that held the spoofing device to the back of the van,” Natalie said.
“Could this be traced back?” Martin asked and turned the piece in his hand.