Autonomous Planetary Landing Methods With Active Olfactory Source Localization

The organic plumes of the solar system and the implied ongoing internal activities represent the most promising targets for astrobiological explorations.Evidence of life,if it exists,may be ejected into space via the plumes emanating from the surface fissures,thus these plumes are regarded as the priority targets for future deep-space missions.The plume vents deposited with fresh organic materials offer an enticing route for direct access to acquire pristine astrobiological samples–this places a premium on the landing site as close to the vent source as possible.However,unlike traditional landing missions where the location of the landing target is known a priori,landers for plume exploration missions are required to locate the landing target –the plume source –autonomously during its descent.We propose the autonomous landing system with the active olfactory source localization for plenary plume explorations.We establish a probabilistic model for plume source localization,given the fluid mechanics of the plume.We also analyze the limitations of the method based on the hidden Markov model and particle filtering in the scenario of planetary plume source localization.The plume source localization problem could be regarded as a nonlinear fitting problem by introducing the probabilistic plume model.The limitations of the traditional method,including overfitting and negative likelihoods are also discussed based on the probabilistic plume model.We propose a sequential Monte Carlo-based source localization method with a biomimetic plume search behaviour for Gaussian plumes.The proposed method overcomes the overfitting and negative likelihood issues by introducing a convolution procedure and then estimates the source position accurately via the sequential Monte Carlo method with a residual degradation procedure.By implementing the proposed biomimetic plume search behaviour,an Enceladus penetrator could search the plume environment by lateral spiral maneuvers and locate the vent source autonomously during its descent.The penetrator could also impact into the icy shell directly using its kinetic with a modest? cost.This offers the prospect of targeting the vent source of the plume for direct access to subsurface material prior to its ejection with a biomimetic behaviour imposing an intermediate ? cost.We propose a novel plume source localization method by tracing the concentration gradient,and the corresponding gradient estimation,transformation and optimization methods for Gaussian plume.By introducing the negative logarithmic concentration gradient,the source position in the direction of normal wind speed could be given as a closed-form solution.The plume source in the wind speed direction could be searched by gradient descent method in the proposed method.In conjunction with the landing guidance law,the proposed method could guide a Mars lander landing at the methane vent source with a sub-optimal trajectory,even the source position is unknown a priori.The energy consumption and landing error are similar to the situation where the landing target is known.We propose a Gaussian mixture plume model and a novel Bayesian source likelihood map method for a scenario of time-varying plume field.The time-varying plume is modelled by a Gaussian mixture function on the basis of the fluid mechanics and the hidden Markov model.We finally propose a novel Bayesian source likelihood map method by introducing a priori distribution for the observation and a Laplace approximation procedure.The observation priori provides a traceable posterior distribution,resolving the overfitting issue caused by the binary concentration value,while the Laplace method calculates the expectation of the source location using a nonlinear least square approach,overcoming the obstacle of the non-Gaussian density convolution.This novel algorithm isolates the intractable segments from the Bayesian procedure and iterate between hyperparameter fitting and Bayesian estimation by the Laplace method.Simulations of Mars methane plume scenario illustrate that the proposed method could locate the plume source efficiently,and could guide the lander landing on the methane vent source with modest fuel consumption and a narrow landing error.The above researches,including the source localization methods and the plume environment search strategies,permit accurate targeting of a planetary plume vent source via the active olfaction autonomously at a modest energy cost.This provides the perspective for autonomous planetary plume exploration missions in the future.