Drones Tech Digest - May 2017

Radar for drones

Echodyne, a radar vision company, has developed a radar that is small enough to be mounted onto a small or medium sized drone. The MESA-DAA uses Echodyne’s metamaterial electronically scanning array (MESA) technology to allow for scanning of the environment both vertically and horizontally, with built-in ability to scan and track simultaneously. Echodyne says that it can spot a CESSNA 140 from 3km away and a Phantom 4 drone from 750m. The device is about double the size of an iPhone 6. 

Safe landing systems for drones

According to Fortune.com, NASA scientists have developed a technology to help drones spot safe places when crash landing. The software has been tested in eight test flights and has landed drones in safe landing zones such as a swamp or drainage ditch. The research is intended to make it safer for people as drones are more widely deployed in commercial and industrial settings such as inspecting powerlines or delivering goods. The software monitors the drone’s status while in flight and if it detects anything breaking down will put the drone into crash landing mode, check a preinstalled set of safe landing zones and then pilot itself towards one so that it can make a safe landing. The scientists plan to carry out further testing of the system over the summer of 2017.  

SkyX pipeline monitoring system

SkyX, an Israeli inspection drone company, has received USD5 million from Chinese Kuang-Chi Group. SkyX uses fixed wing vertical take-off and landing SkyOne drones to carry out pipeline asset monitoring for gas and oil companies. The SkyOne drone can stop on charging stations distributed along the pipeline it is inspecting to recharge – it is capable of flying 100km before needing to be recharged, SkyX claims. Furthermore the drones are claimed to be able to fly in adverse weather conditions. 

NASA traffic management system for drones

NASA invited press to see its Unmanned Aircraft System (UAS) Traffic Management (UTM) system at four FAA test sites in the USA. The tests are focussed on assessing the ability of drones to be piloted beyond line of sight with the aim of improving the UTM technology prototypes. The tests are meant to assess the drones’ abilities to maintain flight plans in windy conditions with radar, cellular signals, ADS-B flight data and GPS provided by the UAS ground control station to the UTM system. This data provided the team with insight into the reliability, delay and accuracy of the drones’ location reporting systems.  

Drones for boats

Wilhelmsen, a Norwegian company providing services to the maritime industry, plans to launch a pilot project using drones to deliver essentials to boats at anchorage in 2017. Currently, to send essentials such as critical documents, medical supplies or cash to a ship at anchorage requires launch boats. Wilhelmsen believes that using drones to deliver the goods will cut costs from an average of USD1500 to just USD150 per trip.

Teaching a drone to fly by crashing

Carnegie Mellon University researchers have taught a drone to fly by allowing it to crash as many times as it needed to – which totalled 11,500 collisions over 40 hours of flight time. The drone equipped with machine learning systems started at a random location in an indoor space and flew until it crashed. After this the drone would try again. The drone has a 30 Hz camera attached that records images, the images are then fed through deep neural networks that evaluate what actions were safe and what actions caused a crash. After the 11,500 crashes the algorithm was able to autonomously navigate the drone in an indoor environment including difficult terrain such as narrow hallways, glass doors and featureless white walls. The drone achieved this by real-time analysis of the video stream to judge what objects and manoeuvres could cause collisions, based on previous situations. 

Quicker drones

Massachusetts Institute of Technology researchers have developed an algorithm that could enable drones with cameras to fly quicker and more nimbly. A factor limiting drone speed is the speed with which cameras on the drone are able to process the images, with 30 miles per hour being the upper limit before lags in processing cause problems. The algorithm developed by the MIT scientists works with a Dynamic Vision Sensor (DVS) developed by a team in Zurich that analyses its environment by taking microsecond interval measurements of brightness. The algorithm enables the camera to tune into only relevant data within the scene. The team believes that the next step is to combine the DVS camera with a conventional camera allowing for a more detailed analysis of what the drone is seeing. 

Autonomous control system for boats

Sea Machines, a US based startup, has developed an autonomous control system for use by USVs or boats up to 24 metres long. The company plans to begin trials of its beta product in the USA and Europe. The SMR 3X system allows for remote control of an equipped boat from onboard another boat, in line of sight, or from over the horizon (non line of sight). 3X-equipped vessels can be tasked to follow another vessel, and the system has the capacity to have 300 equipped vessels working with manned vessels. Radar and infrared cameras are used for collision avoidance, and unmanned surveys can be carried out. SMR 3X can be used in autonomous or remote piloted mode and can be retrofitted into almost any vessel. 

Farming handless

Hands Free Hectare is a project by Harper Adams University in the UK that aims to take a crop from seed to harvest using only autonomous vehicles. In a recent development, the team has drilled a whole hectare site and planted barley seeds in six hours. Prior to this the ground had been sprayed with herbicide also by the autonomous tractor using a GPS-controlled precision sprayer. The tractor uses a drone autopilot system and GPS waymarkers to navigate the field. The waypointing included a signal for the tractor to lift and lower its drill when it reached field edges. The project’s next step is to use its ground rover to capture visuals, and drones to capture multispectral imagery to identify crop emergence. 

Package delivery over long distance

According to the Nevada Institute for Autonomous Systems (NIAS) a new record has been set for the longest long-distance drone delivery. The Nevada UAS Consortium (Team Roadrunner) flew a fixed-wing Unmanned Aerial Vehicle (UAV) over 97 miles (156km) to Austin, Texas. The vehicle flew along a planned route through the National Airspace System (NAS) using mobile command and control, with visual observers along the route that were carrying enhanced radios and cell phone devices that enabled the drone to be flown using cellular technology. 

Add this: