Fail-Safe Navigation for Autonomous Urban Multicopter Flight

Abstract

Current multicopter fail-safe protocols respond with pre-programmed behaviors that aim to minimize possible damage to the aircraft or avoid risk to the nearby population. Existing fail-safe protocols for multicopters involve either returning to a pre-designated home point or executing an automatic landing protocol. Such strategies are rigid and can even increase risk in cases with complex terrain or high overflown population densities. This paper examines three alternative fail-safe protocols for autonomous urban multicopter flight that trade the simplicity of existing protocols with responsiveness to terrain, multicopter state, and overflown population. We utilize nontraditional data and sensor fusion to make an informed landing site selection with a focus on building rooftops as safe urban landing sites. The first protocol assigns a "Clearance to Land" flag to all buildings to determine the nearest rooftop in the area that can serve as a feasible landing site. The second protocol first determines a possible landing site within the multicopter's reachable footprint using lot types provided by a property database. During traversal to that site the multicopter examines the rooftops of overflown buildings and generates an alternative landing plan if it discovers a viable rooftop not previously mapped. The third protocol utilizes the known building locations and their approximate geometries as specified by the database to generate a spanning tree coverage solution followed by the multicopter to search for a viable landing site. A feature-based map of Manhattan generated from the PLUTO and MapPLUTO tax lot databases is used for simulations of all three protocols.

Automatic Classification of Roof Shapes for Multicopter Emergency Landing Site Selection

“…The option of freezing initial layers is exploited with a variable number of frozen layers chosen. When layer 11 is said to be frozen, this means all previous layers, (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11), are frozen during training. All base architectures and tested hyper-parameters are shown in Table 1.…”

“…Small UAS are envisioned to operate at low altitudes, below the airspace occupied by manned aircraft. In a congested environment, the low-flying UAS will operate near buildings thus will have little time or space to execute a "safe ditching" (emergency landing) given an unrecoverable anomaly or failure [2]. Concise and accurate information of nearby landing sites, including building rooftops, is needed in these situations.…”

Multiobjective Modular Strategic Planning Framework for Low-Altitude Missions Within the Urban Air Mobility Ecosystem

“…Landing site definition is a widely investigated problem in the open literature, where several strategies have been developed for correctly tailoring landing site selection to environment topography and ground risk [13], [14]. Besides optimally selecting ground landing site, other approaches have been considered which only assume as landing opportunities vertiport locations [15].…”

Urban Metric Maps for Small Unmanned Aircraft Systems Motion Planning

“…Motion planning algorithms have been used for a wide range of Unmanned Aircraft Systems (UAS) applications including search & rescue [2,3], reconnaissance missions [4,5], sense & avoid [6,7], and navigation through unmapped or uncertain environments [8][9][10]. Motion planners can complement onboard sensor suites to aid in conflict resolution [11] and alternative fail-safe protocols [12] for urban flight. Baseline flight plans are computed and approved prior to flight, but real-time planning may be required to effectively respond to changes in the mission, environment, and/or vehicle performance (e.g., system degradation, failure).…”