Unmanned systems have changed military thinking, but mostly they operate out of sight of the common soldier, loitering at 25,000 ft. or higher, monitoring the battlefield and occasionally reaching out to strike a target.

That is changing as the U.S. military experiments with unmanned cargo air and ground vehicles, bringing ground forces into closer contact with autonomous systems and potentially making them more dependent on unmanned technology.

Teams tackling the disposal of improvised explosive devices have come to rely on their backpack-sized PackBots, but the latest machines are different beasts—whether it is a K-Max unmanned helicopter lifting a 4,000-lb. slung load to a remote outpost or a driverless all-terrain vehicle accompanying a patrol and hauling 1,200 lb. of gear troops would otherwise carry on their backs.

Now the Defense Advance Research Projects Agency wants to take the next step, literally, and develop legged robots that can follow troops anywhere. The bots would be able to: respond to spoken commands, like a trained animal; enter dangerous or compromised environments; and operate available tools and vehicles designed for humans.

Two Lockheed Martin/Kaman K-Max helicopters have logged more than 380 flight hours and carried more than 750,000 lb. of cargo to resupply U.S. Marine Corps outposts since beginning operations in Afghanistan in mid-December. The initial six-month deployment is aimed at demonstrating how unmanned cargo aircraft can take vehicle convoys off dangerous roads.

The helicopters are flying autonomously to two forward locations from a main operating base, carrying food, water, ammunition and other supplies. Loads have averaged 2,500-3,000 lb., but have included a 4,200-lb. generator, says Jim Naylor, director of business development for aviation systems at Lockheed Martin Mission Systems & Sensors.

“The aircraft is controlled from the main operating base and flies autonomously to the forward operating base, where it can land or drop the load,” he says. While the K-Max can complete the entire mission autonomously, “in most cases someone controls the aircraft [at the forward base] and nudges it to where they want the load dropped.” The Navy has also begun retrograde operations, bringing loads back from the forward bases.

One aircraft flies at a time and together they are averaging five flights a day and 95% availability, reaching six a day in March when the mission-capable rate was 100%. Maintenance manhours are averaging less than 1.2/flight hour, dropping to 0.6 in March. “We are demonstrating how low-cost and easy it is to maintain the K-Max,” he says. “We are demonstrating the need for the capability, and that K-Max is the right platform. It's a tough mission, but repetitive lift is what this aircraft was designed for.”

As a next step in unmanned resupply, Lockheed and Kaman are using a third K-Max in the U.S. as a testbed to develop “platform-agnostic” capabilities to help the cargo mission under the U.S. Army's Autonomous Technologies for Unmanned Air Systems (Atuas) joint concept technology demonstration. “Some of these we could move over to the Marine Corps contract,” says Naylor.

Atuas will demonstrate a small beacon that can be placed on the ground at the delivery site. The aircraft will autonomously find the beacon, sense its direction and put down the load a pre-set distance away, avoiding the need for a soldier to take control of the air vehicle at the remote drop site. The program will also demonstrate a ladar-based delivery-site selection system. “No one will need to be there,” says Keith Arthur, team lead for teaming and intelligent systems in the systems integration division of the Army's Aviation Applied Technology Directorate. “They will give it GPS coordinates, the system will scan the area and choose the spot,” he says.

Lockheed and Kaman have also bid for the Office of Naval Research's Autonomous Aerial Cargo/Utility System (Aacus) program, with contract awards expected this month. Aacus will develop a capability enabling unmanned vertical-takeoff-and-landing aircraft to drop off and pick up loads in adverse weather and harsh terrain. The platform-agnostic system will allow the aircraft to autonomously avoid obstacles, select an unprepared landing site and touch down precisely, with the ability to react to unplanned events. The system will communicate with ground personnel, who will be able to negotiate a desired landing site.

Already, during their first weeks of operation in Afghanistan, four robotic cargo vehicles built by Lockheed for the Army have been used to resupply outposts and carry loads ranging from radios and batteries to construction and demolition equipment. The six-wheeled Squad Mission Support Systems (SMSS) arrived in-theater in mid-January for a four-month military utility assessment. At 11 ft. long, they are the largest unmanned ground vehicles deployed in Afghanistan.

The Army's Rapid Equipping Force, through the Robotics Technology Consortium, awarded Lockheed Martin a $975,000 contract to deploy the SMSS to Afghanistan for evaluation. A heavily modified commercial all-terrain vehicle, the diesel-powered SMSS can carry a 1,200-lb. payload for 125 mi. on one tank of fuel. The vehicle also can provide portable power for soldiers, recharging up to 14 batteries simultaneously.

The size and terrain capability of the SMSS allows it to accompany foot patrols and carry their heavy equipment long distances, Lockheed says, but without the need for extra personnel to control the driverless vehicle. Using a “follow-me” feature commanded via hand controller, the SMSS paints a 3-D image of a person using its onboard laser radar and follows that individual to the exclusion of others in the field of view, avoiding the need for a soldier to carry a beacon.

Although the SMSS has the ability to navigate autonomously, for the Afghanistan evaluation the system uses a “bread-crumb-dropping” feature to mark GPS locations every 10 meters (33 ft.) to store routes for reuse. Lockheed has also prototyped voice command, but says this feature was not required on the vehicles deployed to Afghanistan.

Although maneuverable, the SMSS could not follow foot patrols across all terrains, so Darpa is developing a more mobile cargo-carrying robot, the Legged Squad Support System (LS3). Developed by a team led by Boston Dynamics, the prototype robotic “pack mule” completed its first outdoor tests in January and an 18-month platform-refinement test cycle is to begin this summer with Marine and Army involvement.

The four-legged LS3 was developed from Boston Dynamics' BigDog rough-terrain robot, which demonstrated its ability to walk across rubble, climb slopes up to 35 deg. and reach speeds up to 4 mph. Where BigDog could transport a 340-lb. load and travel 12.8 mi. without stopping or refueling, the goal for the larger LS3 is to carry 400 lb. of payload 20 mi. in 24 hr. without refueling, following a squad through rugged terrain and interacting with troops in a way similar to a trained animal and its handler.

LS3 has vision sensors to track a specific individual or autonomously avoid obstacles and Darpa plans to demonstrate a suite of autonomy settings, including: leader-follower tight, where the LS3 attempts to follow the path a soldier takes; leader-follower corridor, in which the vehicle has freedom to make local path decisions; and go-to-waypoint, where the system uses its computer vision to avoid obstacles en route to a designated GPS coordinate. Voice command will also be added during development. Ultimately, Boston Dynamics will build three prototypes, which will participate in Marine Corps exercises.

Darpa is continuing to push ground-robot mobility and capability. Boston Dynamics has demonstrated Cheetah, a four-legged robot that has reached a speed of 18 mph on the laboratory treadmill, flexing its articulated spine like an animal to increase its stride and speed. Tests of a free-running prototype are planned for this year. Now the defense research agency has launched the Robotics Challenge, offering a $2 million prize for designing a robot capable of supervised autonomous response to a simulated disaster.

The 18-month competition will involve a series of events in which the robot will be required to drive a utility vehicle to the disaster site, get out and cross rubble, remove debris blocking a doorway, enter a building, climb a ladder and traverse a walkway, break through a concrete panel, locate and close a valve near a leaking pipe and replace a cooling pump.

The goal of the challenge is to improve the ability of robots to operate in chaotic conditions following a disaster and make use of vehicles and tools available in cities to clear rubble or make repairs. “This challenge is going to test supervised autonomy in perception and decision-making, mounted and dismounted mobility, dexterity, strength and endurance in an environment designed for humans but degraded due to a disaster,” says Gill Pratt, Darpa program manager.