Autonomous Systems & Trusted Autonomy

The wheeled vehicle carried a figure that kept pointing in a fixed direction no matter how the chariot turned, achieved through mechanical gearing that compensated automatically rather than through magnetism, an early example of a machine maintaining a set output without ongoing human correction.

The device automatically adjusted steam flow to keep the engine running at a constant speed regardless of load, an early and influential example of a self-correcting feedback mechanism that engineers still cite as the ancestor of modern automatic control.

His gyroscopic device could hold an aircraft steady in level flight without a pilot's hands on the controls, a demonstration that opened the long path toward machines flying, driving and steering themselves with little human intervention.

The earliest machines regarded as autonomous robots were built in the late 1940s. They established the principle that a machine could act on sensory input without direct human control, foreshadowing later autonomous navigation systems.

Their simplified model of how neurons might compute laid theoretical groundwork that, after decades of limited practical use, became the basis for the deep-learning systems now used for perception and decision-making in autonomous systems.

Drawing on wartime work predicting the movement of enemy aircraft for gun-aiming systems, Wiener set out a general theory of feedback and self-regulation that underlies every subsequent control system for autonomous machinery.

His paper framed the question of machine intelligence in terms of observable behaviour rather than internal mechanism, an approach that continues to shape how autonomy and artificial intelligence are evaluated and debated.

The Kalman filter provided an efficient way to combine noisy sensor readings into a reliable estimate of a system's true state, and was soon adopted in the Apollo guidance computer; it remains embedded in the navigation software of aircraft, missiles and self-driving vehicles.

Early autonomous mobile robots navigated by following buried or laid wires, a highly constrained method suited to structured environments. This marked the first practical deployment of machine navigation without continuous human steering.

HelpMate and the CyberMotion guard robot were the first commercially successful robots able to navigate buildings without human control. They demonstrated that autonomous mobility could support real service tasks in hospitals and security patrols.

From 1990 the competition set full autonomy as the demonstrated objective for aerial robots. It provided a sustained proving ground for autonomous flight, navigation and mission execution.

The result was among the first widely publicised demonstrations of a machine outperforming a leading human expert in a complex decision-making task, a symbolic marker later invoked whenever autonomous systems crossed a new capability threshold.

Navigation moved beyond fixed wire guidance to triangulation from beacons placed in the environment. This loosened the dependence on physical infrastructure embedded in floors and widened the range of deployable settings.

The challenges spurred rapid development of autonomous driving capability for ground vehicles. They drew university and industry teams into the field and seeded much of the later self-driving car industry.

A flood of small autonomous vacuum cleaners reached consumers from 2002, opening the domestic robot market. It was the first mass-market demonstration that autonomous navigation could be made cheap and reliable enough for household use.

These machines could build their own laser-based maps of a building and navigate open areas as well as corridors. Simultaneous mapping and localisation removed the need for pre-installed guidance infrastructure.

The vehicle, named Stanley, completed a course through the Mojave Desert without a driver after every entrant had failed to finish a similar race the year before, a result widely credited with launching the modern self-driving vehicle industry.

Its performance recognising objects in photographs far exceeded previous techniques and persuaded much of the field to shift toward neural-network approaches, providing the perception backbone that autonomous vehicles, drones and robots rely on to interpret sensor data today.

The competition aimed to produce the first autonomous robot capable of operating on oil and gas production sites. It extended autonomy into hazardous industrial inspection environments.

In the final Sample Return Robot Centennial Challenge a rover demonstrated autonomous navigation, decision-making, and sample detection, retrieval and return. It was a benchmark for end-to-end autonomy without human intervention in unstructured terrain.

The company started charging passengers for rides within a geofenced area of the city, its first paid autonomous ride-hailing operation, though a human safety monitor remained in the front seat and the service area stayed far smaller than rival operators.

The partnership was intended to let riders in Asian and Middle Eastern markets book Apollo Go vehicles through Uber, a route to international expansion that sidestepped the years Western operators had spent building their own driverless fleets from scratch.

Software that ranks and recommends targets from large volumes of intelligence data remained central to the pace of strikes in Gaza, drawing sustained criticism from rights groups and some allied governments over the degree of human oversight applied to machine-generated recommendations.

Quadruped and wheeled autonomous carriers were tested for moving ammunition and supplies across difficult terrain near the border with China, reducing the need for soldiers or pack animals to carry loads manually at extreme altitude.

The programme worked to move autonomy software and small uncrewed systems from prototype into units' hands at scale, with firms such as Anduril supplying command-and-control software intended to let a small number of operators supervise many machines at once.

The unit, focused on integrating drones and ground robots with artificial intelligence for reconnaissance and strike, grew its role in exercises as the military sought to offset a shrinking pool of conscripts with autonomous and semi-autonomous systems.

The Alphabet subsidiary added highway driving and further metropolitan areas to its paid service while beginning exploratory testing in Tokyo, extending a lead built on years of accumulated autonomous driving data over rivals still confined to limited zones.

First-person-view attack drones increasingly used onboard computer vision to lock onto a target in the last stretch of flight, allowing the weapon to keep flying and strike accurately even after electronic warfare severed the radio link to its operator.

The emirate's roads authority added testing and rollout agreements covering robotaxi and autonomous shuttle operators, working toward a stated target of making around a quarter of journeys autonomous by 2030 as part of a wider push to position the city as an autonomy testbed.

The two Chinese autonomous driving companies completed initial public offerings within weeks of each other, giving investors direct exposure to China's robotaxi and autonomous trucking sector despite continued political friction over Chinese technology in the West.

The robot autonomously connects to and charges an electric vehicle, and had been piloted at Hyundai Motor Group's headquarters since 2021. It exemplifies autonomy applied to routine physical service tasks in infrastructure.

Astro extended domestic autonomy beyond cleaning into security, monitoring and eldercare roles. Its launch signalled major technology firms entering the household autonomous robot market.

Meeting under the Convention on Certain Conventional Weapons, the expert group examined the ethical concerns of robots wielding weapons. It was the principal international forum addressing constraints on autonomy in armed systems.

The Ripsaw M5 is an unmanned ground vehicle capable of over 40 mph with an 8,000 lb payload. Four prototypes were to be delivered for company-level trials in late 2021 to assess integrating unmanned vehicles into ground combat.

The smaller companion robot taught children to code with support for Python, Blockly and Raspberry Pi. It represented the diffusion of humanoid robot technology into consumer education products.

By 2018 the humanoid Sophia combined scripting software, a chat system and the OpenCog AI system for general reasoning. It illustrated the integration of conversational AI with embodied autonomous platforms.

The city moved to restrict rather than encourage sidewalk robot testing by imposing a permit requirement. It marked an early instance of municipal pushback against autonomous systems in public space.

The bills allowed autonomous delivery robots to travel on sidewalks and use crosswalks statewide from 1 July 2017. Virginia thereby created one of the earliest statewide legal regimes for autonomous ground delivery.

The bill authorised a pilot programme for delivery robots on public ways, later scheduled from September to December 2017. It was among the first legislative frameworks permitting autonomous machines to operate in shared pedestrian space.