-
On the Road with 16 Neurons: Mental Imagery with Bio-inspired Deep Neural Networks
Authors:
Alice Plebe,
Mauro Da Lio
Abstract:
This paper proposes a strategy for visual prediction in the context of autonomous driving. Humans, when not distracted or drunk, are still the best drivers you can currently find. For this reason we take inspiration from two theoretical ideas about the human mind and its neural organization. The first idea concerns how the brain uses a hierarchical structure of neuron ensembles to extract abstract…
▽ More
This paper proposes a strategy for visual prediction in the context of autonomous driving. Humans, when not distracted or drunk, are still the best drivers you can currently find. For this reason we take inspiration from two theoretical ideas about the human mind and its neural organization. The first idea concerns how the brain uses a hierarchical structure of neuron ensembles to extract abstract concepts from visual experience and code them into compact representations. The second idea suggests that these neural perceptual representations are not neutral but functional to the prediction of the future state of affairs in the environment. Similarly, the prediction mechanism is not neutral but oriented to the current planning of a future action. We identify within the deep learning framework two artificial counterparts of the aforementioned neurocognitive theories. We find a correspondence between the first theoretical idea and the architecture of convolutional autoencoders, while we translate the second theory into a training procedure that learns compact representations which are not neutral but oriented to driving tasks, from two distinct perspectives. From a static perspective, we force groups of neural units in the compact representations to distinctly represent specific concepts crucial to the driving task. From a dynamic perspective, we encourage the compact representations to be predictive of how the current road scenario will change in the future. We successfully learn compact representations that use as few as 16 neural units for each of the two basic driving concepts we consider: car and lane. We prove the efficiency of our proposed perceptual representations on the SYNTHIA dataset. Our source code is available at https://github.com/3lis/rnn_vae
△ Less
Submitted 9 March, 2020;
originally announced March 2020.
-
Target-distractor Synchrony Affects Performance in a Novel Motor Task for Studying Action Selection
Authors:
Sebastian James,
Olivia A. Bell,
Muhammed A. M. Nazli,
Rachel E. Pearce,
Jonathan Spencer,
Katie Tyrrell,
Phillip J. Paine,
Timothy J. Heaton,
Sean Anderson,
Mauro Da Lio,
Kevin Gurney
Abstract:
The study of action selection in humans can present challenges of task design since our actions are usually defined by many degrees of freedom and therefore occupy a large action-space. While saccadic eye-movement offers a more constrained paradigm for investigating action selection, the study of reach-and-grasp in upper limbs has often been defined by more complex scenarios, not easily interpreta…
▽ More
The study of action selection in humans can present challenges of task design since our actions are usually defined by many degrees of freedom and therefore occupy a large action-space. While saccadic eye-movement offers a more constrained paradigm for investigating action selection, the study of reach-and-grasp in upper limbs has often been defined by more complex scenarios, not easily interpretable in terms of such selection. Here we present a novel motor behaviour task which addresses this by limiting the action space to a single degree of freedom in which subjects have to track (using a stylus) a vertical coloured target line displayed on a tablet computer, whilst ignoring a similarly oriented distractor line in a different colour. We ran this task with 55 subjects and showed that, in agreement with previous studies, the presence of the distractor generally increases the movement latency and directional error rate. Further, we used two distractor conditions according to whether the distractor's location changes asynchronously or synchronously with the location of the target. We found that the asynchronous distractor yielded poorer performance than its synchronous counterpart, with significantly higher movement latencies and higher error rates. We interpret these results in an action selection framework with two actions (move left or right) and competing 'action requests' offered by the target and distractor. As such, the results provide insights into action selection performance in humans and supply data for directly constraining future computational models therein.
△ Less
Submitted 3 May, 2017;
originally announced May 2017.
-
Global asymptotic stability of a PID control system with Coulomb friction
Authors:
Andrea Bisoffi,
Mauro Da Lio,
Andrew R. Teel,
Luca Zaccarian
Abstract:
We propose a model for representing a point mass subject to Coulomb friction in feedback with a PID controller, based on a differential inclusion comprising all the possible magnitudes of static friction during the stick phase. For this model we study the set of all equilibria and we establish its global asymptotic stability using a discontinuous Lyapunov-like function, and a suitable LaSalle's in…
▽ More
We propose a model for representing a point mass subject to Coulomb friction in feedback with a PID controller, based on a differential inclusion comprising all the possible magnitudes of static friction during the stick phase. For this model we study the set of all equilibria and we establish its global asymptotic stability using a discontinuous Lyapunov-like function, and a suitable LaSalle's invariance principle. We finally use well-posedness of the proposed model to establish useful robustness results, including an ISS property from a suitable input in a perturbed context. Simulation results are also given to illustrate our statements.
△ Less
Submitted 28 September, 2016;
originally announced September 2016.
-
Global results on reset-induced periodic trajectories of planar systems
Authors:
Andrea Bisoffi,
Fulvio Forni,
Mauro Da Lio,
Luca Zaccarian
Abstract:
We study the existence of asymptotically stable periodic trajectories induced by reset feedback. The analysis is developed for a planar system. Casting the problem into the hybrid setting, we show that a periodic orbit arises from the balance between the energy dissipated during flows and the energy restored by resets, at jumps. The stability of the periodic orbit is studied with hybrid Lyapunov t…
▽ More
We study the existence of asymptotically stable periodic trajectories induced by reset feedback. The analysis is developed for a planar system. Casting the problem into the hybrid setting, we show that a periodic orbit arises from the balance between the energy dissipated during flows and the energy restored by resets, at jumps. The stability of the periodic orbit is studied with hybrid Lyapunov tools. The satisfaction of the so-called hybrid basic conditions ensures the robustness of the asymptotic stability. Extensions of the approach to more general mechanical systems are discussed.
△ Less
Submitted 23 September, 2015;
originally announced September 2015.
-
Measuring random force noise for LISA aboard the LISA Pathfinder mission
Authors:
D. Bortoluzzi,
L. Carbone,
A. Cavalleri,
M. Da Lio,
R. Dolesi,
C. D. Hoyle,
M. Hueller,
S. Vitale,
W. J. Weber
Abstract:
The LTP (LISA Testflight Package), to be flown aboard the ESA / NASA LISA Pathfinder mission, aims to demonstrate drag-free control for LISA test masses with acceleration noise below 30 fm/s^2/Hz^1/2 from 1-30 mHz. This paper describes the LTP measurement of random, position independent forces acting on the test masses. In addition to putting an overall upper limit for all source of random force…
▽ More
The LTP (LISA Testflight Package), to be flown aboard the ESA / NASA LISA Pathfinder mission, aims to demonstrate drag-free control for LISA test masses with acceleration noise below 30 fm/s^2/Hz^1/2 from 1-30 mHz. This paper describes the LTP measurement of random, position independent forces acting on the test masses. In addition to putting an overall upper limit for all source of random force noise, LTP will measure the conversion of several key disturbances into acceleration noise and thus allow a more detailed characterization of the drag-free performance to be expected for LISA.
△ Less
Submitted 4 February, 2004;
originally announced February 2004.
