Research-Number1.png
 

VISION & PERCEPTION

Use-inspired research challenges addressed by FAIR-SPACE:

The potential of 3D sensing in space robotics is largely untapped and is yet to be fully realised. Stereo-vision based depth perception using optical cameras is the de facto standard, and in comparison, space-rated LIDAR is more power hungry and bulky although it brings advantages such as range and sensing robustness needed by future missions.

Recognition of human-made objects (e.g. sample caches and tools, other spacecraft), natural hazards/landmarks, and dynamic events such as weather phenomena (e.g. Martian dust devils) are important for complex tasks such as in-space robotic servicing, safer space navigation, cache acquisition for sample returns, and opportunistic scientific observation of events. These tasks will require the use of higher-level representations, going beyond point clouds and boundary representations, to include solid objects and shapes. These are also likely to play a central role in future implementations of fast mapping.

Space sensing and perception are crucial for providing rapid, autonomous navigation and manipulation for future orbital and planetary missions, and are therefore directly linked to Guidance, Navigation and Control (GNC). Simultaneous Mapping and Localization (SLAM) being the state-of-the-art, fast mapping method which has not been widely adopted in space applications. The method possesses myriad parameters that need tuning to allow effective use in a given scenario. These include thresholds that control feature-matching, RANSAC parameters, and criteria to decide when to introduce new map elements or trigger a search for loop closure matches.

Initial Research Projects

  • Sparse 3D sensing and map fusion: Develop RGB-D based sparse sampling and 3D map construction fusing with optical camera data under sparsity constraints.

  • Fusion of event-based cameras and depth sensing: Hardware and software development to allow fusion of event-based/range sensors and detection of Region of Interest.

  • Resilient SLAM for space: Develop low computational, robust simultaneous mapping and navigation algorithms for space conditions.

RGB-D like sensing in space / Courtesy of Surrey

RGB-D like sensing in space / Courtesy of Surrey

theme1b.jpg
theme1a.jpg

Research-Number2.png
 

Mobility & Mechanisms

Use-inspired research challenges addressed by FAIR-SPACE:

Future space missions require manned and unmanned planetary vehicles to access and traverse extreme terrain topographies (such as steep and deep craters, gullies, canyons, lava tubes, and soft, friable terrains) in faster traverse speed, longer range and greater coverage. The current monomorphic surface vehicle designs (such as Mars rovers) are limited.

There are growing demands for increased/enhanced dexterity for robotic manipulation in space including sample manipulation, assembly of large structures, moving heavy tools, or preparing a site prior to crew arrival. Current drawbacks in relevant technologies relate to limitations in material and mechatronic properties of space-rated manipulator components, control of high DOF, real-time computation and reaction to external forces, power requirements, and lack of multi-modal control systems.

Technologies in grappling, capture & sampling mechanisms are needed to increase handling capabilities of space debris and in-orbit assets, and sample acquisition and handling capabilities for sample returns and in-situ resource utilization. The space sector currently lacks several mechanism solutions, such as: lightweight, deep drilling under low gravity; sampling at depth without losing or contaminating the samples; and grappling and capture of non-cooperative targets (without capture aids, in a wide range of masses or dynamic spin/free drift).

Initial Research Projects

  • Rough terrain multi-contact locomotion: develop all-terrain mobility on rugged surfaces using actuator-optimized mechanism with coordinated manipulation for path clearance.

  • Robust manipulation and grappling: develop novel dexterous robotic arm/hand, and its optimal control to coordinate whole system’s redundancy for energy-efficient manipulation tasks.

  • Subsurface sample acquisition: develop bio-inspired robust mechanisms for drilling/sample acquisition suitable for metamorphic deployment from the surface vehicles.

Gripper for non-cooperative target Courtesy of Salford

Gripper for non-cooperative target
Courtesy of Salford

Mobile chassis for dynamic locomotion  Courtesy of Surrey

Mobile chassis for dynamic locomotion
Courtesy of Surrey

Lightweight, low power “wasp drill” Courtesy of Surrey

Lightweight, low power “wasp drill”
Courtesy of Surrey


Research-Number3.png
 

AI & Autonomy

Use-inspired research challenges addressed by FAIR-SPACE:

GNC algorithms have traditionally put a premium on positioning and pose estimation quality at a cost to runtime performance. Space robotics & autonomous systems (RAS) require low-latency and resource-efficient computation to observe their immediate environment and perform tasks within it.

The decision-making software of future space RAS is expected to offer a full range of autonomy levels from remote operation, semi-autonomy to fully autonomous operation (i.e. autonomy level E1 to E4 according to European Cooperation for Space Standardization or ECSS). Typical spacecraft are currently only capable of autonomy up to level E3. Next generation missions require advanced on-board software to fully automate decision making and task planning in space RAS.

There is a growing demand for longer-duration operations, and survivability in a context of attacks, degradations or failures of individual components or subsystems while communication with the Earth is limited. If reliable software redundancy techniques could be implemented, redundant hardware can be scaled back to significantly reduce weight and cost.

Initial Research Projects

  • Resource-aware reconfigurable GNC: Design high-fidelity GNC system that can adjust its computational load depending on the sensing and computational resources available and can tolerate faults.

  • Multi-objective task planning: Develop dynamic coordination of both global and local tasks subject to the external operating environment, resources, and mission time, on the basis of a set of performance metrics (e.g. energy, distance, time).

Manipulation GNC of cooperative target from pseudo-fixed or free-flying platform in space / Courtesy of Surrey

Manipulation GNC of cooperative target from pseudo-fixed or free-flying platform in space / Courtesy of Surrey

Cooperative GNC of manipulators / Courtesy of Edinburgh

Cooperative GNC of manipulators / Courtesy of Edinburgh

FAIR-SPACE in-space experiment for AI & autonomy / Courtesy of Surrey

FAIR-SPACE in-space experiment for AI & autonomy / Courtesy of Surrey


Research-Number4.png
 

Astronaut-Robot Interaction

Use-inspired research challenges addressed by FAIR-SPACE:

GNC algorithms have traditionally put a premium on positioning and pose estimation quality at a cost to runtime performance. Space robotics & autonomous systems (RAS) require low-latency and resource-efficient computation to observe their immediate environment and perform tasks within it.

The decision-making software of future space RAS is expected to offer a full range of autonomy levels from remote operation, semi-autonomy to fully autonomous operation (i.e. autonomy level E1 to E4 according to European Cooperation for Space Standardization or ECSS). Typical spacecraft are currently only capable of autonomy up to level E3. Next generation missions require advanced on-board software to fully automate decision making and task planning in space RAS.

There is a growing demand for longer-duration operations, and survivability in a context of attacks, degradations or failures of individual components or subsystems while communication with the Earth is limited. If reliable software redundancy techniques could be implemented, redundant hardware can be scaled back to significantly reduce weight and cost.

Initial Research Projects

  • Wearable astronaut-robot proximal interaction technology: Create a human-robot interface technology that can be used to control space robotics during physically demanding scenarios such as during a spacewalk or whilst performing inter-vehicular maintenance at ISS.

  • Combined Augmented Reality/Virtual Reality for space environments: Create a combined AR/VR methodology that enables astronauts to efficiently interact with robots, and respective system prototypes. AR will be applied to enhance the information content available through the human vision system, robot-mounted cameras and VR scenes.

Humanoid Valkyrie robot / Courtesy of Edinburgh

Humanoid Valkyrie robot / Courtesy of Edinburgh

Astronaut wearable technologies / Courtesy of Imperial

Astronaut wearable technologies / Courtesy of Imperial

theme4c.jpg

Research-Number5.png
 

System Engineering

Use-inspired research challenges addressed by FAIR-SPACE:

GNC algorithms have traditionally put a premium on positioning and pose estimation quality at a cost to runtime performance. Space robotics & autonomous systems (RAS) require low-latency and resource-efficient computation to observe their immediate environment and perform tasks within it.

The decision-making software of future space RAS is expected to offer a full range of autonomy levels from remote operation, semi-autonomy to fully autonomous operation (i.e. autonomy level E1 to E4 according to European Cooperation for Space Standardization or ECSS). Typical spacecraft are currently only capable of autonomy up to level E3. Next generation missions require advanced on-board software to fully automate decision making and task planning in space RAS.

There is a growing demand for longer-duration operations, and survivability in a context of attacks, degradations or failures of individual components or subsystems while communication with the Earth is limited. If reliable software redundancy techniques could be implemented, redundant hardware can be scaled back to significantly reduce weight and cost.

Initial Research Projects

  • Modularity and Verifiability: Map out library of key components and architectural designs for space RAS that are amenable to a range of V&V, prioritizing the three use-cases.

  • Threat Landscape for Space RAS: Undertake an analysis of the cyber-security landscape for space RAS and design protocols for future standardization.

  • Dimensions for Verification: Research into formalisation and logical properties for key space verification dimensions (communications, spatial, time, energy, etc), and their combination into a scalable, flexible and applicable range of verification tools.

Verifiable autonomy architecture / Courtesy of Liverpool

Verifiable autonomy architecture / Courtesy of Liverpool

theme5b.jpg