AUTONEWS

Northeastern Seattle’s custom Ford Mustang lets students conduct research from the driver’s seat

It was a long way to the car seat as we know it today. Early motorists lacked individual adjustment options for their seats. It took many decades until comfort and safety features such as headrests or easy-to-use longitudinal adjustments found their way into series production. Nowadays there are a lot of high-tech car seats, up to twelve percent of the production costs of a vehicle are in the seats. Initially, the body and seats are more reminiscent of a carriage than a car.

The seating position is extremely upright and high. Of course, there are no adjustment options and despite the noble leather upholstery, the driver is sitting on a soft board. Anyone who had to drive this vehicle for more than half an hour definitely got off with back pain. As time progressed, the seats continued to consist of a solidly fixed bench, but the seating position shifted significantly. In early rudimentary cars, occupants sat on chairs with feet placed directly on the floor, whereas in cars from the 1930s, the seats were fixed, resembling the configuration seen in modern vehicles today. However, there are still no adjustment options, and we searched in vain for the thigh support.

Cars from the 50s already offer a more comfortable seating position. Now the seats can also be adjusted lengthwise. Of course, this method is rudimentary and complicated, but it works. This construction allows drivers and passengers of larger and smaller sizes to travel comfortably. In addition, it is also possible to fold the backrests of the front seats forward, facilitating easier access for rear-seat passengers. But there is no sign of a headrest or side support here, and comfort takes its toll because the adjustable seat frame must be considered quite critically when it comes to occupant safety.

By the end of the 70s, thanks to the extendable telescopic rod, the height of the seats could be adjusted with relatively low effort. Furthermore, during this period, an optional headrest is already being used, the first sign that safety is becoming a development criterion alongside comfort. By the 90s, the length, height and angle of the seats were adjustable in all directions. They offer good lateral support, sufficient thigh support and, of course, the headrest which is nowadays legally required.

Today, a driver's seat with multiple options weighs 35–40 kg and is a complex system consisting of a metal frame, upholstery, covers, various hydraulic shock absorbers, electric motors, heating filaments, fans, cables, and connectors. A modern seat consists of a total of around 500 parts. After the drivetrain, the driver's seat is one of the most expensive components in the vehicle. Manufacturers usually offer their cars to their customers with basic seats, which can be upgraded with options such as massage or heating and ventilating functions.

Seats are important for safe driving. Ergonomic seats play a key role in minimizing physical strain and fatigue, thereby supporting sustained attention and concentration. This helps prevent rapid decreases in focus, consequently aiding in accident prevention. In modern vehicles, an airbag is installed on the side of the backrest towards the body, the purpose of which is to reduce side collisions. The far-side airbag is a novelty in vehicle development. This airbag is designed to prevent the driver and front passenger from colliding in the event of a side impact by deploying from the seat and inflating towards the headliner. This new type of airbag protects those in front rows from injuries due to interaction between passengers. These brand-new concepts can prevent about one-third of serious injuries related to the side collision. In 2020, the first series featuring the utilization of far-side airbags in a passenger car was observed.

 A tall garage gate on Northeastern University’s Seattle campus rolled up to reveal the snarling headlights of a Ford Mustang Mach-E wrapped in the signature Husky red-and-white color scheme.

Behind the wheel of the sleek, all-electric SUV was Xian Li, a professor of electrical and computer engineering and instructor of the Robotics Sensing and Navigation course. Though Li admitted to being an aggressive driver in his own car, he steered the Mustang past the nearby downtown headquarters of Amazon, Microsoft and Meta with careful precision. After all, he said, there was precious cargo to protect.

Equipped with an array of cameras, navigation sensors, and an onboard computer, the Mustang serves as a sporty, drivable data-collection tool for students and faculty on Northeastern’s Seattle campus. The car soaks up images and navigation data at a clip of 10 gigabytes per minute. It provides information that Li said is ripe for analysis and experimentation.

The nearly 5,000-pound vehicle is one of the campus’s most unusual research platforms, exposing students interested in robotics, computer vision and autonomous driving to hardware constraints, operational challenges and environmental factors that cannot be captured in simulations alone. Li encourages them to use the data collected from the Mustang to build and test their own navigation algorithms.

“Not all students have access to something like this,” said Xinlei Li, a master’s student in electrical and computer engineering who helped Li mount several of the Mustang’s sensors in the spring of 2026. “I always wondered whether I would ever drive something like this, so this is pretty cool to me.”

Autonomous driving technology allows vehicles to draw from their surrounding environment to make informed movements. The technology helps vehicles with safe driving behavior such as staying within painted freeway lanes, using road signs to identify the proper speed limit, or recognizing a crossing pedestrian in time to stop. Different kinds of onboard cameras and sensors collect data on traffic conditions, obstacles and the movement of nearby vehicles.

Northeastern’s Mustang is equipped with four cameras, a LiDAR scanner to generate three-dimensional maps, radar to measure the speed and distance of other objects, a GPS system to track location, an inertial measure unit (IMU) to calculate acceleration, altitude and other movements, and eight terabytes of onboard storage. The various components are strategically bolted to the car’s roof, strapped to the interior console and headrests or secured in the trunk.

“The more data you get, the more surroundings that the robot will understand, so that they can perform better in the real world,” said Xinlei Li, who had previously collected sensor data as part of his co-op with Kentucky-based moving walkway manufacturer Beltways. The challenging part is understanding the strengths and weaknesses of each sensor type, and learning how to combine them to generate the clearest picture of the environment, he said.

Self-driving cars, like those from Tesla, already boast powerful autonomous driving systems, but the domain still has a long way to go. Li said Tesla’s vehicles and their advanced “Full Self-Driving (Supervised)” capabilities have only reached Level 2 on a six-level scale of self-driving sophistication, from Level 0 to Level 5, according to a taxonomy published by SAE International, a professional association of engineers. A vehicle at Level 5 would require no intervention or thought on the part of the passenger, Li said, meaning they could safely fall asleep during long-distance drives. 

The electric car soaks up images and navigation data at a clip of 10 gigabytes per minute. Photo by Alyssa Stone/Northeastern University(Northeastern University)

Vehicles with Level 5 capabilities currently do not exist, nor will they in the next decade, Li said. Their eventual consumer cost is anyone’s guess, he said, as the legislation and regulations to allow for complete self-driving would need to be established before any such vehicle hits the market.

The Ford Mustang Mach-E is not manufactured with full self-driving capabilities. Instead, it has Ford BlueCruise, an SAE Level 2 hands-free system that keeps the car within highway lanes and manages cruise control as long as drivers keep their eyes on the road. 

But Li and his students are working to hack the vehicle and implement more advanced autonomous features – an effort to bring the car closer to SAE Level 3, he said. By integrating and customizing certain algorithms in the controlling parts of the car, such as its steering and thruster, Li hopes he and his students can one day make the Mustang fully self-driving.

The lessons learned from sensor data collection can also be applied outside the personal mobility space, he said.

For example, Xinlei Li – who is not related to Xian Li – said he has used navigation data to develop technology that can help the visually impaired move around in unfamiliar spaces. 

Vittoria Bitton, who is pursuing a master’s degree in computer vision and machine learning at Northeastern Seattle, researched the ways that environmental imagery could aid search-and-rescue operations in areas with low visibility.

Bitton arrived at Northeastern with nearly a decade of professional experience, having worked with computer vision pipelines and secured a senior software engineer position with electric vehicle manufacturer Rivian before her December graduation. In Li’s class, she acts as a mentor to students at earlier stages of their careers. 

Bitton’s contributions include explaining the technical aspects of the Mustang’s sensors and suggesting components to add for future cohorts’ use – such as an onboard neural processor, which uses a machine learning model to recognize complex patterns.

She recommended that Li and his team think about using AI to create a perception system inside the vehicle, which would use the onboard cameras and sensors to interpret and build a real-time three-dimensional rendering of the surrounding environment.

Another idea is “hacking the car so that students can understand the architecture of an EV,” she said after Li’s class met inside the garage. 

Students and faculty at Northeastern University in Seattle have transformed a Ford Mustang Mach-E into a high-tech mobile research platform. Led by Professor Xian Li, the initiative uses the all-electric SUV as a laboratory on wheels for a course on Robotics Sensing and Navigation.

The vehicle, wrapped in the university’s signature red and white color scheme, travels the streets of Seattle, collecting vast amounts of environmental data.

Technical details and functionality:

• Massive data collection: The car captures images and navigation information at an impressive rate of 10 gigabytes per minute.

• Integrated sensors: The Mustang is equipped with a comprehensive suite of tools, including LiDAR, radar, advanced cameras, GPS, and inertial sensors.

• Practical objective: Students use this data to analyze and experiment with autonomous driving algorithms. The focus is on enhancing safety-critical behaviors, such as lane-keeping, identifying speed limits via traffic signs, and early pedestrian detection.

• In-depth exploration: Beyond navigation, students work on developing AI-based perception systems to generate real-time 3D renderings of the surroundings, while also exploring the vehicle’s own electrical architecture.

The project stands out for creating a hands-on learning environment where programming errors and real-world testing are integral parts of the robotics innovation process.

Northeastern University